Ultrasound Technology For Clinical Practitioners

Ultrasound Technology
for Clinical Practitioners

Ultrasound
Technology for
Clinical Practitioners
Crispian Oates
Newcastle University, UK
This edition first published 2023
© 2023 John Wiley & Sons Ltd
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system,
or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or
otherwise, except as permitted by law. Advice on how to obtain permission to reuse material from
this title is available at http://www.wiley.com/go/permissions.
The right of Crispian Oates to be identified as the author of this work has been asserted in
accordance with law.
Registered Office
John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, USA
John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ, UK
For details of our global editorial offices, customer services, and more information about Wiley
products visit us at www.wiley.com.
Wiley also publishes its books in a variety of electronic formats and by print-on-demand. Some
content that appears in standard print versions of this book may not be available in other formats.
Trademarks: Wiley and the Wiley logo are trademarks or registered trademarks of John Wiley
& Sons, Inc. and/or its affiliates in the United States and other countries and may not be used
without written permission. All other trademarks are the property of their respective owners.
John Wiley & Sons, Inc. is not associated with any product or vendor mentioned in this book.
Limit of Liability/Disclaimer of Warranty

While the publisher and authors have used their best efforts in preparing this work, they make
no representations or warranties with respect to the accuracy or completeness of the contents
of this work and specifically disclaim all warranties, including without limitation any implied
warranties of merchantability or fitness for a particular purpose. No warranty may be created or
extended by sales representatives, written sales materials or promotional statements for this work.
The fact that an organization, website, or product is referred to in this work as a citation and/or
potential source of further information does not mean that the publisher and authors endorse the
information or services the organization, website, or product may provide or recommendations
it may make. This work is sold with the understanding that the publisher is not engaged in
rendering professional services. The advice and strategies contained herein may not be suitable
for your situation. You should consult with a specialist where appropriate. Further, readers
should be aware that websites listed in this work may have changed or disappeared between
when this work was written and when it is read. Neither the publisher nor authors shall be liable
for any loss of profit or any other commercial damages, including but not limited to special,
incidental, consequential, or other damages.
Library of Congress Cataloging-in-Publication Data Applied for:

ISBN 9781119690825
Cover Image: Courtesy of the Author

Cover Design: Wiley
Set in 10.5/13pt STIXTwoText by Straive, Pondicherry, India

This book is dedicated to Tony Whittingham who has been my guide and mentor in all
things ultrasonic and to Marion my constant companion and encourager.
Contents
Acknowledgments xvii
List of Abbreviations xix
Introduction 1
CHAPTER 1 The Basic Physics of Ultrasound 5

Sound Waves 5
Describing Waves 9
Energy in a Sound Wave 11
Ultrasound Pulses 12
Energy Spectrum of a Pulse 13
Bandwidth 14
Quality Factor 16
Speed of Sound (C) 16
Characteristic Acoustic Impedance, Z0 20
Energy in a Sound Wave 22
Decibels 23
CHAPTER 2 The Interaction of Ultrasound with Tissue 25

Reflection and Transmission at a Plane Interface 25
Specular Reflection 25
Reflection at an Angle and Refraction 27
Poor Visualisation 29
Scattering 30
Total Scattered Power 31
The Speckle Pattern 31
Attenuation 34
Attenuation Coefficient (α) 35
Summary 37
The Journey of the Ultrasound Pulse 37
User Control 37
Transmit Power 37
References 38
vii
viii Contents
CHAPTER 3 Beam Shapes 39

Simple Beam Shape Model 40
Huygen’s Wavelet Model and Diffraction 43
Focusing 44
Beam Forming with Transducer Arrays 47
Grating Lobes 48
Apodization 50
Beam Steering 50
Beam Shape and Beam Steering 52
Electronic Focusing 52
Resolution 54
Axial Resolution 55
Lateral Resolution 55
Resolution Versus Penetration 57
Clutter 58
Reference 58
CHAPTER 4 The Ultrasound Probe 59

The Transducer 59
Piezoelectric Effect 59
Single Crystal Technology 61
Resonant Frequency 62
Backing Layer 62
Matching Layer 63
Front Face Lens 65
Wide Band Transducers 65
Construction of an Array 66
CMUT Technology 66
1-D, 1.5-D, and 2-D Arrays 68
The Hanafy Lens 68
1.5-D Array 70
2-D Array 70
References 72
CHAPTER 5 Image Formation 73

Image Modes 74
A-Mode (Amplitude Mode) 74
M-Mode (Motion Mode) 74
B-Mode or Greyscale Image (Brightness Mode) 76
Linear Image Formation 76
Linear Array 76
Frame Rate 77
Curvilinear Array 77
Phased Array 78
Extended Field of View 78
Panoramic View 78
3D Imaging 80
Hand Scanning 80
Contents ix
Mechanical 3D Probe 80
2D Matrix Array 80
3D Image Display 81
Cine Loop 82
Endoprobes 82
Choosing A Probe 84
Focusing 84
Transmit Focus 84
Receive Focus 84
Increasing Frame Rate 86
Synthetic Aperture Imaging 86
User Control 86
Resolution/Penetration 86
Ultrasound Harmonics 89
Harmonic Imaging 90
Pros and Cons of Harmonic Imaging 91
Coded Excitation 92
References 94
CHAPTER 6 The B-Mode Scanner 95

Transmission Side of a Scanner 95
Beam Former 96
PRF – Pulse Repetition Frequency 96
User Controls 96
Transmit Power 96
Depth of Focus 96
Receive Side of a Scanner (RX) 97
Beam Former 97
Gain 97
User Controls – Gain 98
Time Gain Compensation 98
Analogue to Digital Converter – ADC 99
Advantages of Digitising 101
Echo Detection 101
Dynamic Range and Transfer Function (Greyscale Mapping) 102
Dynamic Range 103
Transfer Function or Grey Scale Mapping 103
Contrast Resolution 106
User Controls 106
Image Memory 106
Frame Freeze 106
Read and Write Zoom 107
Read Zoom 107
Write Zoom 108
Image Processing 108
User Control 108
Exam Presets 108
Measurements 109
x Contents
CHAPTER 7 Image Quality and Artefacts 111

Acoustic Window 111
Frame Rate: Frames Per Second (fps) 112
Interlacing Scan Lines 113
Interpolation – Writing in ‘Extra Lines’ 114
Speckle 115
Frame Averaging or Persistence 116
User Control 117
Spatial Compound Imaging 117
Adaptive Filtering 118
Edge Enhancement 118
Noise in an Image 119
Colour Enhancement of Greyscale Images 120
Human Visual Perception 120
How to Set Up a Monitor 121
Image Flicker 121
Ambient Lighting 121
Artefacts 122
Assumptions 122
Speed of Sound Artefacts 122
Refraction 122
Axial Misplacement 123
Attenuation Artefacts 127
Poorly Adjusted Time Gain Control (TGC) 127
Acoustic Shadowing 127
Post Cystic Enhancement 128
Shadowing from the Transducer Surface 129
Reflection Artefacts 130
Mirror Image Artefact 130
Reverberation 131
Reverberation Artefact 131
Probe – Interface Reverberation 131
Ringdown 131
Comet Tail or Ringing 132
Anisotropy 134
Beam Shape Artefacts 135
Slice Thickness Artefact 135
Side Lobes and Grating Lobes 136
Temporal Artefacts 137
Image Blurring 139
Final Example 139
References 140
CHAPTER 8 Principles of Doppler Ultrasound 141

The Doppler Effect 141
The Doppler Equation 143
Duplex Ultrasound 144
Contents xi
CW Doppler 145
Block Diagram of CW Doppler Instrument 146
Detection of Direction of Flow 146
The Doppler Waveform Display (Sonogram) 148
Examples 148
1. Slowly Moving Blood Flow in a Vessel 148
2. Fast Moving Blood Flow in a Vessel 149
CW Doppler Summary 152
Pulsed Wave Doppler (PW Doppler) and Range Gating 152
Detection of the Doppler Signal 153
Block Diagram of Pulsed Doppler Instrument
with Sonogram Display 154
Aliasing 154
Appearance of Aliasing on the Sonogram 156
Observing Low Velocities 158
PW Doppler Summary 159
The PW Doppler Pulses 159
Intrinsic Spectral Broadening (ISB) 160
Question: What Doppler Angle Should We Use? 162
User Controls 163
Angle Correction 163
Doppler Beam Steering 163
Length and Depth of Sample Volume 163
Scale 164
Sweep Speed 164
Baseline Offset and Invert 165
Doppler Transmit Power 165
Doppler Gain 165
Peak Velocity Envelope 165
Autotrace 166
Average Velocity 167
Doppler Artefacts 170
Aliasing 170
Intrinsic Spectral Broadening 170
Wall Thump 171
User Control 172
Waveform Ghosting 172
Shadowing 173
References 173
CHAPTER 9 Principles of Colour Doppler Ultrasound 175

Autocorrelation 177
Colour Scale 180
Frame Rate 181
User Controls 181
Colour Box 181
Frame Averaging 182
Sampling 182
xii Contents
CDU and the Doppler Angle 183

Colour Aliasing 183
User Controls 185
Scale 185
Invert 186
Discrimination of Stationary Targets 187
Imaging Small Vessels 187
User Controls 188
Wall Filter 188
B-Mode Priority 188
Power Doppler (PD) 188
CDU Artefacts 190
Aliasing 190
Shadowing 190
Wall Discrimination 192
Flash Artefact 192
Mirror Image 192
Twinkle Artefact 192
Temporal Artefact 192
Colour Sensitivity 192
Presets 194
Colour M-Mode 194
Tissue Doppler Imaging (TDI) 194
Myocardial Strain Imaging 197
Strain 197
Strain Rate (SR) 197
Using TDI to Measure Strain 198
Speckle Tracking Echocardiography STE 199
STE Display 201
References 202
CHAPTER 10 Making Measurements 203

Accuracy 204
Precision 204
How Accurate or Precise Do We Need To Be? 205
Reproducibility 205
Systematic and Random Errors 206
Random Errors 206
Ultrasound Measurements in Practice 206
Physical Constraints 207
Scaling 207
Caliper Accuracy 207
Image Resolution 208
Sonographer-Based Constraints 209
Principles for Making Reliable Measurements 209
Target Visualisation 209
Frame Freeze 209
Use First Interface 209
Contents xiii
Measurements Protocols 209

Actual Measurements 211
Factoring Variability 211
Ultrafine Measurements 212
Measurement of Circumference, Area, and Volume 213
Direct Measurement 213
Assume a Simple Shape 213
Use a Simplified Model of the Target 214
Slicing Models 214
Surface Segmentation and Automated Methods 214
Doppler Waveform Measurements 216
Image/Sonogram Size 216
Scale 216
Gain 217
Freeze the Image 217
Weak Signals 217
Doppler Angle 217
Waveform Indices 219
Pulsatility Index (PI) 219
Resistance Index (RI) 220
A/B Ratio 220
Colour Doppler Ultrasound 221
Measurement of Volume Flow Q 221
Cross-Section Area A 222
References 224
CHAPTER 11 Safety and Quality Assurance 225

Energy, Power, and Intensity 226
Measuring Intensity 227
Intensity 227
Factors Affecting Damage Potential 230
Thermal Effects 231
Thermal Index (TI) 232
TIS 233
TIB 233
TIC 233
Transducer Self-Heating 234
Nonthermal Effects 235
Radiation Force 235
Streaming 235
Cavitation 236
Mechanical Index (MI) 239
ALARA 239
Contrast Agents 240
Clinical Risk 240
Bio-effects from Ultrasound 240
Quality Assurance and Routine Checks 241
xiv Contents
Suggested Routine User Checks 241

Annual Service 243
The Use of Test Objects 244
Personal Risk Management 245
WRRSI Risk Mitigation 245
In Summary 246
New Techniques in Ultrasound 246
References 247
CHAPTER 12 Advanced Topics 249

Contrast Agents (CA) 249
Behaviour of Bubbles in the Ultrasound Field 251
Contrast Agent Harmonics 252
Amplitude Modulation 253
Pulse Inversion 254
Flashing 254
Advanced Micro-Bubble Techniques 255
B-Flow Blood Vessel Imaging 256
Doppler Measurement of Pressure Gradients 260
Advanced Image Processing 261
Artificial Intelligence 261
Segmentation 262
Examples (1–3) 262
Computer-Aided Diagnosis (CAD) 263
Feature Identification 266
Examples 266
Classification 266
Deep Learning 266
Diagnosis with Cad 268
Examples of CAD 269
Fusion Imaging 269
Needle Visualisation and Guidance 271
References 274
CHAPTER 13 Ultrafast Ultrasound 277

Synthetic Aperture Imaging (SA) 278
Plane-Wave Beamforming 279
Summary 283
Speed of Sound Correction 283
Benefits 286
Ultrafast Doppler 286
Ultrafast Colour Doppler (UFCD) 286
Microvascular Imaging 289
Pulsed Wave Doppler (PW) 290
Vector Flow Imaging (VFI) 291
The Flow Vector 291
Directional Beam Forming 292
Contents xv
Transverse Oscillation (TO) 293

Display of Flow Vectors 296
Vector Display 296
Colour Display 296
References 298
CHAPTER 14 Elastography 301

Background Theory 302
Elastography 303
Methods of Applying The Distorting Force 303
Strain Elastography (SE) 303
SE with Manual or Physiological Palpation 303
User Controls 307
Frequency 307
Frame Rate 307
Tracking Window 307
Frame Reject 307
Noise Reject 307
Line Density 308
Frame Averaging 308
Dynamic Range 308
Colour Map 308
Colour Blend 308
SE Artefacts 310
Stress Concentration Artefacts 312
Edge Enhancement Effects 312
Appearance of Cysts 313
Blue/Green/Red Artefact 314
Shadowing 314
Acoustic Radiation Force Impulse Imaging (ARFI Imaging) 314
Pros and Cons of ARFI 315
Strain Ratio 316
Shear Wave Elastography (SWE) 316
Background Theory 316
Point SWE (PSWE) 320
Supersonic Shear Imaging (SSI) 322
Shear Wave Compounding 323
SWE Artefacts 325
Bang Artefact 325
References 326
Appendix 1: Knobology 329

Appendix 2: Handling Equations and Decibels 335
Appendix 3: The Unfocused Transducer Beam Shape 345
Index 349
Acknowledgements
In writing this book, I have been helped by many friends and colleagues. I am very
grateful for all their contributions and help, and indebted to them for their advice.
It began with interaction with my students over many years and a set of honed lec-
ture notes. More specifically, I have been particularly helped by Gareth Bolton at
the University of Cumbria who very kindly let me use his facilities and scanners to
get a significant number of the clinical images demonstrating machine settings. He
also gave me valuable feedback, reading drafts and looking at the material from the
student-teacher point of view. Stephen Klarich and Jamie Wild gave me invaluable
advice on the practicalities of using elastography and Chris Eggett advised me in rela-
tion to echocardiography. Barry Ward advised on quality assurance. I am also thank-
ful to Kathia Fiaschi, Carmel Moran, and Heather Venables. Those mentioned and
others kindly provided images as indicated throughout the book. Putting together
a book like this takes a considerable time and Covid lockdown certainly helped in
providing that time. So too did my family and particularly my wife Marion. She has
been a great support and encourager and I am deeply thankful to her. However, hav-
ing acknowledged those who have helped and contributed in various ways, the work
is mine as are any errors and mistakes that remain within it.
Crispian Oates
April 2020
xvii
List of Abbreviations
ADC analogue to digital converter

AI artificial intelligence
A-mode amplitude mode
ARFI acoustic radiation force impulse
B-flow B-mode flow
B-mode brightness mode
CA contrast agent
CAD computer aided diagnosis
CCA common carotid artery
CDU colour Doppler ultrasound (CFM)
CEUS contrast enhance ultrasound
CFM colour flow mapping (CDU)
CLA curvilinear array
CMUT capacitative micromachined ultrasound transducer
CPU central processing unit
CT computerised tomography
CUTE computed ultrasound tomography in echo mode
CW continuous wave
DGC depth gain control (TGC)
ECG electrocardiogram
FFT fast Fourier transform
FPS frames per second
FR frame rate
FWHM full width at half maximum (beamwidth)
GPU graphics processing unit
ICA internal carotid artery
ISATA spatial average temporal average intensity
ISB intrinsic spectral broadening
ISPPA spatial peak, peak average intensity
xix
xx List of Abbreviations
ISPTA spatial peak temporal average intensity

ISPTP spatial peak temporal peak intensity
IUCD intra-uterine contraceptive device
LA linear array
MI mechanical index
M-mode motion mode
MRI magnetic resonance imaging
PA phased array
PD power Doppler
PI pulsatility index
PRF pulse repetition frequency
PSV peak systolic velocity
pSWE point shear wave elastography
PVDF polyvinylidene flouride
PWD pulse wave Doppler
PZT lead zirconate titanate
QA quality assurance
RBC red blood cell
RF radio frequency
RI resistance index
ROI region of interest
RSI repetitive strain injury
Rx receive (signal)
SA synthetic aperture
SCA subclavian artery
SE strain elastography
SNR signal to noise ratio
SoS speed of sound
SR strain rate
SRT systolic rise time
SSI supersonic shear (wave) imaging
STE speckle tracking echocardiography
SV sample volume
SWE shear wave elastography
TDI tissue Doppler imaging
TGC time gain control (DGC)
TI thermal index
TIB thermal index for bone in view
TIC thermal index for superficial bone in view
List of Abbreviations xxi
TIS thermal index for soft tissue

Tx transmit (signal)
UFCD ultrafast colour Doppler
UFUS ultrafast ultrasound
VFI vector flow imaging
WRRSI work related repetitive strain injury
Introduction
This book covers the essential physics and technology of diagnostic ultrasound
needed by someone practicing ultrasound in the clinical setting, with ultrasound as a
primary or significant component of their job. For simplicity, the term ‘sonographer’
has been used throughout for this person but ultrasound is used by a wide range of
personnel in clinical practice including doctors, echocardiographers, vascular scien-
tists, midwives, nurse practitioners, and physiotherapists. The book is designed to be
accessible to all of these practitioners. Each chapter is liberally illustrated with easily
reproducible drawings and clinical images to demonstrate the point being made. The
use of equations has been kept to a minimum. Where used, equations are useful in
showing the relationship between one factor and another and where changing one
thing can have clinical or safety implications. The term ‘scanner’ refers to the ultra-
sound machine.
Over the years, ultrasound machines have become more user friendly and the
machine performs many functions without the user being aware of what is being
changed, for example the use of presets for particular patient examinations. It is
important to have an understanding of what your equipment, being applied to a
patient, is doing. By knowing more about the technology behind the scanner, you will
increase in confidence in handling your scanner and be able to be assured that you are
obtaining the optimum diagnostic information and are operating in a safe manner for
yourself and the patient. The patient may also be reassured that they are being treated
by a competent practitioner who knows their equipment and what they are doing.
The emphasis throughout will be on what the user needs to know in order to
drive the ultrasound scanner correctly and effectively in order to obtain the best
images. We will also look at the technical factors that must be taken into account
when interpreting the images to make clinical judgments. In a number of places, it
will be necessary or useful to explain a point in greater detail or add additional but less
essential information. These will be indicated by using green shaded boxes. Where a
point of specific relevance to daily practice is made, green text is used. Key terms are
highlighted in bold type.
Ultrasound uses sound waves of a higher pitch than the audible range to form
images from within the body. The ultrasound is produced by a transducer probe that
is typically placed on the skin, after it has had a liquid gel applied to it. Short pulses
Ultrasound Technology for Clinical Practitioners, First Edition. Crispian Oates.

© 2023 John Wiley & Sons Ltd. Published 2023 by John Wiley & Sons Ltd.
1
2 Introduction
of ultrasound are transmitted into the body and are reflected, forming echoes that in
turn are picked up by the ultrasound probe. The received signal is then processed to
form the image we see. This method of forming images is called the pulse-echo tech-
nique and is similar to that used by sonar on boats or by radar.
In Figure 0.1, we see three typical ultrasound greyscale images. Looking at the
three images, the first thing to notice is that there are two basic image formats. A rec-
tilinear format or linear scan (a) and a sector shaped format or sector scan (b,c).
Both formats show a cross-sectional slice through the body as though the body had
been cut open, going deep from the transducer probe at the skin surface, and we are
looking down onto the cut surface (Figure 0.2).
As shown, the transducer/skin surface is at the top of the linear image and at
the narrower point of the two sector images. The orientation of the image may be
chosen by the user. The three images are produced by different ultrasound probes and
(a) (b)
(c)
FIGURE 0.1 Three typical greyscale or B-mode images from (a) a linear array, (b) a curvilinear
array and (c) a phased array. Image (b) shows the greyscale used down the left-hand side.
Introduction 3
(a) (b)
FIGURE 0.2 The image plane within the body (a) and the ultrasound image seen (b).
are useful for different clinical examinations. In the first image (a), the linear scan is
produced by a linear array probe and is useful for looking at small parts including
musculoskeletal examinations and vascular work. The second image (b) is produced
by a curvilinear probe and is useful for situations where an extended target is being
viewed, for example the abdomen or a foetus. The third image (c) is produced by
a phased array probe. This probe has a small footprint on the skin but is able to
show an extended field of view in the body, for example viewing the heart from bet-
ween the ribs.
At its most basic, the process of image formation can be thought of as using a
narrow beam of ultrasound to sweep through the tissue across the image plane or
scan plane, like we might sweep a torch beam across a dark room, to build up a pic-
ture of what is there. The image is therefore built up from a series of lines transmitted
out from the transducer and going deep into the body, laid side-by-side to form the
image. Each line uses the pulse-echo principle to receive echoes from along that line.
In the case of the linear scan, it is as though the beam was swept in a straight line along
the surface of the skin across the width of the probe. In the case of the phased array, it is
as though the probe was rocked at one point on the skin to sweep the beam in a sector.
If we know how fast the pulse travels through the tissue, we can time the echoes
coming back and so calibrate the depth of the echo targets in centimetres away from
the probe. The marks down the side of each image indicate depth from the trans-
ducer probe.
Looking at the images, details of structures producing echoes are shown as bright
and dark marks matching a greyscale as seen on the left side of Figure 0.1b. This
type of image is known as a B-mode or greyscale image. Some parts of the image
are clearer and more obvious than other parts. In order to interpret these images,
the scanner must firstly be set up to produce the optimum quality image and sec-
ondly, the person interpreting needs to understand what is really being imaged and
what is artefact or misrepresented in the image. As for all imaging modalities, the
4 Introduction
sonographer needs anatomical knowledge and an understanding of how the images

are formed and what limitations the technique has in order to correctly interpret
the images.
Whilst modern machines have automated many of the processes involved in
ultrasound imaging, there remain a large number of variables under user control that
the sonographer must manage effectively to produce optimal images and maximise
the diagnostic potential of ultrasound.
We begin our look at ultrasound technology by considering what ultrasound
is and how it interacts with tissue. We then move on to look at the production of a
B-mode or greyscale image and its interpretation before going on to consider Doppler
ultrasound. We then look at making measurements, safety of ultrasound, and quality
assurance before moving on to advanced topics and the latest developments with ultra-
fast techniques. Finally, we look at elastography. Three appendices cover a check list
for performing a scan that covers ‘knobology’, the basic manipulation of equations,
and a detailed look at the ultrasound beam.
CHAPTER 1
The Basic Physics
of Ultrasound
SOUND WAVES
A sound wave is a fluctuating variation in pressure within a medium such as air,

water, or solid material. Our ears are sensitive to such pressure changes in air, and we
hear sounds all the time. The faster the changes in pressure take place, the higher the
pitch or frequency of the sound we hear. Frequency is measured in hertz (Hz) and,
for a young person, their hearing goes from 20 Hz to 20 kHz. Middle C on a piano is
261 Hz. A sound above 20 kHz is called ultrasound (‘beyond sound’), in other words,
we cannot hear it. Dogs can hear sounds of higher frequency than we can, and bats
use ultrasound, up to 200 kHz, for echo location in the dark. The ultrasound we use
for medical imaging is in the range of megahertz (MHz), far above anything we can
hear. As we will see, the reason for going to such high frequencies is that we can then
make narrow beams of ultrasound that we can point in a particular direction and
which can produce high-resolution images showing fine details.
We can think of a medium like air or water as a collection of molecules with mass
connected by springs that represent the forces between the molecules (Figure 1.1).
If you push on one molecule, it will move closer to the adjacent molecule and exert
a force so that it too begins to move. Adjacent molecules having been squashed
together will then repel one another and recover to their resting position. They will
keep moving beyond their resting position due to their momentum. The force holding
them together then becomes an attractive force that pulls the molecules back towards

5
6 Chapter 1 The Basic Physics of Ultrasound
Intermolecular
forces
M M M M
V– V+
Particle velocity
Displacement
FIGURE 1.1 Ball and spring model of a sound wave travelling in some medium.
their resting position again as they are then stretched apart. That is, each molecule
will move forwards then backwards in the direction of the applied force. This oscil-
lating molecular motion within a material is the basis of a sound wave.
DEFINITION
A sound wave is a longitudinal pressure wave travelling at the speed of sound
through a medium (e.g. air, water, and soft tissue).
The pressure is the force exerted from the excess density (number of molecules
per unit volume) of molecules above or below the average density of the medium as
the molecules get alternately squashed and pulled apart from one another. In other
words, it is the excess pressure about the mean resting pressure.
A sound wave may be generated by a piston moving forwards and backwards.
This is what a normal loudspeaker does (Figure 1.2). It has a diaphragm that moves
forwards and backwards driven by an electric signal. This will alternately push, then
move away, from the molecules in front of it. These molecules will then alternately
push on to those in front of them and then pull on them, and so on, as described
Oscillating diaphragm
FIGURE 1.2 A loudspeaker driven by an electric signal acts as a piston on the air in front of it.
Sound Waves 7
P + – + – + – + – + – + –
C
λ
FIGURE 1.3 Oscillating piston producing a longitudinal pressure wave, a sound wave consisting
of compression (+) followed by rarefaction (−). c is the speed of sound, and λ is the wavelength.
above. This will create a series of compressions and rarefactions in the molecules of
the medium that move away from the source of movement (the piston), i.e. a sound
wave as shown in Figure 1.3.
The speed at which they move away from the source is called the speed of sound
(c). The speed of sound can be considered constant for each medium, so the speed of
sound in air is 330 m·s−1, and if a plane goes faster than this it breaks the sound barrier.
The speed of sound in water varies with temperature and at 20 °C is 1480 m·s−1. The
average speed of sound in soft tissue is 1540 m·s−1, and, as will be seen, this is a key
number for ultrasound imaging.
For ultrasound imaging, the ultrasound transducer acts exactly like the loud-
speaker pushing and pulling the molecules of the medium in front of it.
NOTE
The individual molecules oscillate backwards and forwards about a mean position,
but the pressure disturbance (p) propagates forward at the speed of sound (c). It is
the moving disturbance that is the sound wave.
A typical sine wave plot of a sound wave is seen if we plot the change in pressure
(p) at a given point against time, or if we plot the change in pressure versus distance
away from the piston.
Looking at one position in space versus time (t), we see the pressure increasing
and decreasing as the sound wave passes by (Figure 1.4a).
The frequency (f) is the number of cycles (peaks) passing a given point in
one second.
The period (T) is the time taken to complete 1 cycle. The relationship between
period and frequency is
1
f
T
Looking at one instant of time versus distance (x) away from the sound source
(Figure 1.4b), we see the pressure increases and decreases as we move away from the
transducer.
T λ
P+ P+
P t P X
P– P–
(a) (b)
FIGURE 1.4 Illustration of the change in pressure with time at one point in space (a) and the
change in pressure with distance x from the sound source (b).
The wavelength (λ) of the sound is then defined as the distance in space between
two successive peaks on the wave.
THE SPEED OF SOUND EQUATION

The relationship between speed of sound (c), frequency ( f ), and wavelength (λ) is
c f.
Relationship Between Pressure, Particle Velocity and Particle

Displacement
We can also plot the change in particle velocity (v) versus time and the particle
displacement (s), from its resting position, versus time, to give similar graphs
(Figure 1.5).
P+
t Excess
pressure
V–
t Particle
velocity
S+
t Particle
displacement
FIGURE 1.5 The sine waves produced by plotting the change in pressure, particle velocity,
and displacement associated with a sound wave.
Describing Waves 9
Comparing these waves, we see that the pressure is greatest when the particle
velocity is greatest and the particle displacement is greatest when the rising pres-
sure passes through its mean zero level.
When we simply talk about a ‘sound wave,’ we usually mean the (excess) pressure
wave – also known as acoustic pressure. It is acoustic pressure that our ultrasound
transducers are sensitive to and detect.
NOTE
Do not confuse particle velocity with the speed of sound. Particle velocity is
movement at a molecular level, whereas the speed of sound is the speed at which
the sound wave propagates through the medium.
A transducer is anything that converts one form of energy into another form.
The loudspeaker and the ultrasound transducer both convert electrical energy into
sound energy and so are transducers.
NOTES
• As a sound wave propagates its frequency remains constant at the same fre-
quency as the transducer ‘piston’ oscillates.
• For a given medium, the speed of sound is constant (it may vary with tem-
perature). This means that if the transmitted frequency increases, the wave-
length must get shorter to balance the speed of the sound equation.
• Key Concept: ‘High frequencies give short wavelengths’.
DESCRIBING WAVES
The amplitude of a wave is the difference between the peak value and the mean
zero value. We can also define the peak-to-peak amplitude A+ to A− as shown in
Figure 1.6.
A+
A–
FIGURE 1.6 Definition of wave amplitude A.

FIGURE 1.7 Illustration of a spinning bicycle wheel over a moving roll of paper.
180° 270° 360° = 0°

0° 90°
FIGURE 1.8 Definition of the phase angle of a sound wave.
The phase of a wave is a point along the course of one period of the wave
expressed as an angle.
Think of a spinning bicycle wheel mounted above a moving roll of paper as shown
in Figure 1.7. On the paper, we mark the vertical distance of the valve away from the
hub at each moment as the wheel spins round. What results is a sine wave drawn
on the paper. By the time, the wheel has gone round once, you would have drawn
one period of the sine wave and the valve would have travelled round 360°. So, by
measuring the angle of the valve as it goes round, we can mark the phase angle along
the sine wave as shown in Figure 1.8. One cycle is equal to 360° (hence, frequency is
equal to ‘cycles’ per second).
This gives us a very useful way to compare two sine waves. If one wave has a phase
angle of 45° at the same time another sine wave has a phase angle of 0°, we know
where the peak of one wave is compared to the other (Figure 1.9). If we know the fre-
quency, amplitude, and phase of a wave, we know everything about it. The wavelength
will depend on the speed of sound of the medium the sound wave travels through.
A special case of sound wave transmission is seen if we look at a single frequency
sound wave propagating in one direction through a uniform medium from an infi-
nitely large sound source. This type of wave is called a plane wave. The wavefronts
Energy In a Sound Wave 11
90 180 270 360
0° 45°
FIGURE 1.9 Two sine waves with a phase difference of 45°.
FIGURE 1.10 A plane wave showing the wavefronts parallel to one another.
(wave peaks) are then parallel to one another (Figure 1.10). Sea waves coming on to a
straight beach closely approximate this situation.
Plane Waves
If we compare the pressure wave with the particle displacement and particle veloc-
ity waves for a plane wave, they all have the same frequency. Looking at Figure 1.5
and comparing the phase difference between them, we find that the excess pres-
sure and particle velocity waves are in phase with each other (i.e. peaks occur at
the same time), whilst the particle displacement wave is 90° ahead of the pressure
wave. That is, its peak occurs a quarter of a cycle before that of the pressure and
particle velocity waves. Their relative amplitudes will depend on the medium they
are travelling through.
ENERGY IN A SOUND WAVE
In order to move the sound source piston backwards and forwards, work must be done
and this requires energy. The disturbance created in the molecules can then make mol-
ecules some distance away from the piston move, so the sound wave must be transport-
ing energy through the medium. The average pressure in a sound wave is zero as the
amplitude oscillates equally above and below the mean. But the amplitude squared
Amp2
Amp 0
0
Energy ∝ (amplitude)2
FIGURE 1.11 The relationship between the amplitude of a sound wave and the energy carried
by the sound wave.
always has a positive value (think of (−1) × (−1) = +1) (Figure 1.11). The energy
carried by a sound wave is proportional to the pressure amplitude squared (amp)2.
So, for the acoustic pressure p, energy E p2 .
The energy carried by a sound wave is important when considering the safety of
ultrasound exposure in the body. The ultrasound is depositing energy in tissue, and it
can cause damage if too much energy is deposited in one place.
We will consider the energy and safety of ultrasound further in Chapter 11.
ULTRASOUND PULSES
A pulse is a very short burst of sound. Most of the imaging we do uses very short
pulses of sound, so we can tell where the echoes are coming from at any one time.
If we consider a pulse in time, it will have a shape and duration. A typical imaging
pulse might have a large amplitude at first then die away (Figure 1.12).
• Peak amplitude – typically, 1 atmosphere = 105 Pa

• Particle velocity – typically, peak amplitude value ~ 10 cm s−1
• Particle displacement – typically, peak amplitude value ~ 20 nm (1 nm = 10−9 m)
• Pulse duration – three—four cycles for B-mode ultrasound at 5 MHz
• Pulse duration – 0.2 μs × 4 = 0.8 μs
We can also consider the shape of the pulse in space at one instant in time.
Note the reversal of wave shape – peak amplitude comes first in time and in space
(Figure 1.13).
• Pulse length in tissue – three—four cycles

• For B-mode ultrasound at 5 MHz in soft tissue
c = 1540 m·s−1 λ = 0.3 mm × 4 to give a pulse length of 1.2 mm
Energy Spectrum of a Pulse 13
Peak amplitude
Pulse duration
FIGURE 1.12 Typical pulse shape of a short pulse of ultrasound.
Pulse envelope
Direction of travel
Pulse length
FIGURE 1.13 The pulse envelope of an ultrasound pulse in space.
We can characterise the shape of the pulse by drawing a line through successive
peaks to show the pulse envelope, as shown by the dashed line.
ENERGY SPECTRUM OF A PULSE
DEFINITION
A spectrum is a plot of energy against frequency.
• For example, in a light spectrum, as seen in a rainbow, red has a lower

f requency and blue has a higher frequency. The energy is seen in the bright-
ness of the colours.
The rate at which energy is produced or transmitted is called power, which is

equal to energy per unit time (‘rate’ means ‘per unit time’).
For a sound wave, the power spectrum is a plot of the power in the sound wave
versus the frequency of the wave. Figure 1.14 shows what the spectrum of a simple
sine wave of frequency f0 looks like.
Constant amplitute
E(f )
f0 free
FIGURE 1.14 The single line power spectrum of a continuous sound wave of a single
frequency, f0.
It has all of its energy (E0) at a single frequency ( f0 ).

For a pulse, we can ask the following questions:
• What is the frequency of a pulse?

• What does the spectrum of a single pulse look like?
BANDWIDTH
Looking at the pulse we saw in Figure 1.12 or the pulse in Figure 1.15, we see that no
two cycles are the same, unlike the pure sine wave where every cycle is identical. On
the other hand, the waveform for most of the pulse itself does look like a sine wave.
So, what is its frequency?
The answer is that the pulse has a spectrum that centres around f0, but extra
frequencies need to be included on either side to account for the sine wave stopping
at the ends of the pulse. Figure 1.16 shows the spectrum of the pulse has a band-
width or range of frequencies centred about f0. We talk about the ‘full width at half
maximum’, or FWHM, as a measure of the bandwidth. This is the width of the spec-
trum, in hertz, measured at the half height of the peak around the centre frequency.
We can then look at the shape a pulse has and we can consider the energy that the
pulse carries. The energy of a pulse becomes an important factor when we look at the
safety of ultrasound.
When talking about the shape of a pulse over time, we talk about the time
domain and when talking about the power spectrum of the pulse, we talk about
the frequency domain. Figure 1.17 shows what the energy spectrum is for various
pulses, i.e. the frequency content of a single pulse.
Notice how the very long pulse looks more like the sine wave than the very short
pulse. Similarly, its spectral bandwidth is narrow and close to that of the sine wave
frequency, f0. The shorter the pulse, the wider the bandwidth of the spectrum, until
with an infinitesimally short pulse the bandwidth is infinite and there is no associated
single frequency, fo. All frequencies are equally present. As an example of an infini-
tesimally short pulse, in order to study how all the frequencies respond at once in a
concert hall, sound engineers use a gunshot from a starting pistol. A sharp hand clap
also has a very wide range of frequencies emitted.
Bandwidth 15
P
Time
FIGURE 1.15 A 7-cycle pulse.
Maximum
FWHM
f0
FIGURE 1.16 The full width at half maximum or FWHM bandwidth of an ultrasound pulse
with centre frequency, f0.
Time domain Frequency domain
Constant amplitute
E(f)
P
Infinitely long pulse
Time (constant frequency sine
f0 freq wave)
E(f)
P FWHM
Time Long pulse
f0 freq
E(f) Short pulse

P
Time
f0 freq
Infinitesimally short pulse

E(f)
(e.g. sound from a starting
pistol or a sharp hand clap)
P
t0 f0 freq
Time
FIGURE 1.17 The relationship between the time domain picture and the frequency domain
picture for various pulse shapes.
SUMMARY
There is a reciprocal relationship between the time domain and the frequency
domain. An infinitely long sine wave has a single frequency whilst a single instan-
taneous sound has infinite bandwidth.
The importance of this relationship for medical ultrasound is that in order to

have very short pulses and to give high-resolution images, we must use transducers
that have a wide bandwidth so they can transmit and receive across a wide range of
frequencies. For this reason, manufacturers will emphasise the fact that they have
wideband transducers. Typical pulse shapes for B-mode and Doppler pulses are
shown in Figure 1.18.
An approximation of the bandwidth of a given pulse length is given by
1
Bandwidth ~ 2
Pulse length
So, a short pulse gives a wider bandwidth. For a typical diagnostic pulse of
three cycles for B-mode, FWHM bandwidth, ≈f0, where f0 is the centre frequency.
For example, a 5 MHz pulse has an FWHM bandwidth of ±2.5 MHz
Quality Factor
Some texts refer to the quality factor Q of a resonator such as an ultrasound trans-
ducer (not to be confused with Q for flow volume!)
Centre frequency, f0
Q
Bandwidth
So, a wideband transducer has a low Q, and a transducer with a narrow

bandwidth has a high Q.
SPEED OF SOUND (C)
(Note: do not confuse the speed of sound with particle velocity!)

Returning to the ball and spring model of a material, Figure 1.19, we see that the
model consists of a series of masses representing the molecules, connected together
by springs of a certain stiffness representing the forces between the molecules. In
5-Dec-12
13:47:49
B:MA(FFT(A))
5 MHz
110 mV
A:Average(1)
.2 μs
1.00 V A
200 swps
.2 μs
1 1 V DC
500 MS/s
2 .5 V AC 1 DC –2.96 V
NORMAL
(a)
5-Dec-12
14:34:59
B:MA(FFT(A))
5 MHz
132 mV
A:Average(1)
.2 μs
0.99 V A
200 swps
B
.2 μs
1 1 V DC
500 MS/s
2 .5 V AC 1 DC –3.62 V
STOPPED
(b)
FIGURE 1.18 (a) A B-mode pulse and (b) a pulse wave Doppler pulse. The lower left corner of
each image shows the spectrum of each pulse. Note that the longer pulse in (b) has a narrower
bandwidth. Source: Courtesy: Jie Tong.
order to get a sound wave to propagate through the medium we need to get a mass
moving and that moving mass must transmit its movement onto the next molecule
through the spring. The lighter the mass, the easier it will be to get it moving, and the
Intermolecular
forces
M M M M
FIGURE 1.19 Ball and spring model of the molecules of a material showing the inter-molecular
forces between the oscillating molecular masses.
FIGURE 1.20 Definition of the bulk modulus showing the decrease in volume with an increase
in applied pressure, P.
stiffer the spring, the faster it will transmit that movement on to the next molecule.
We may, therefore, expect that light stiff materials will have a high speed of sound,
and dense soft materials have a low speed of sound. For example, aluminium, a light
rigid metal, has a speed of sound 5100 m·s−1, whilst in air the molecules are also light
but are very poorly connected, and the speed of sound is only 330 m·s−1.
In terms of what we can measure, the average mass per unit volume of material
is called the density ρ (kg·m−3), and the stiffness is a measure of the forces between
the molecules
For a volume of a material, its resistance to such forces is called the bulk
modulus K.
The bulk modulus relates the fractional change in volume ΔV/V to the applied
pressure ΔP (Figure 1.20).
P 1 2
Bulk modulus, K K V kg m s
V
The minus sign occurs because the volume decreases with an increase in pressure.
K
Speed of sound in a material, c
NOTES
• At a given temperature, the speed of sound is constant.
• Speed of sound is independent of frequency (for most purposes in medical
ultrasound).
• Materials differ more in their stiffness than in their density. Therefore, stiff-
ness is a better guide to predicting c than density.
Looking at Table 1.1, we see that the speeds of sound for different soft tissues
are quite close together. The average speed of sound in soft tissue in the body is
1540 m·s−1 ± 5%. This is close enough to 1540 m·s−1 for us to use this value in most
circumstances in clinical ultrasound. The value 1540 m·s−1 is a key number to
remember when considering how ultrasound propagates through soft tissues.
The variation in speed of sound from 1540 m·s−1 will produce artefacts in the image
(see Chapter 7). However, sometimes we need to apply a more accurate value to our
measurements, for example when making measurements of the eye in ophthalmol-
ogy. In Chapter 13, we will see how such variations in speed of sound may be corrected
to improve image quality.
TABLE 1.1 Showing the Speed of Sound for Various Materials

Speed of sound table
Material Speed of Sound (m·s−1)
Air 330
Water 1480
Plastic (Perspex) 2730
PZT transducer 3741
Fat 1450
Brain 1546 Soft tissue average
Liver 1550
Kidney 1560 1540 ± 5%

Blood 1570
Muscle 1580
Bone (cortical) 3500
CHARACTERISTIC ACOUSTIC IMPEDANCE, Z0
QUESTION
How easily does a molecule in the medium move in response to a given change in
acoustic pressure?
The answer is given in a quantity called the characteristic acoustic impedance,

Z0. This is a measure of how easily a molecule in the medium moves in response to
the excess pressure (p) of the sound wave – assuming the sound wave is a plane wave.
Z0 is constant for a given medium.
Characteristic Acoustic Impedance
p Excess pressure
At a molecular level, acoustic impedance Z 0
v Particle velocity
Recall that for a plane wave, the pressure wave and the particle velocity are in
phase with one another. The ratio of their amplitudes will therefore be constant
at all points in the wave and equal to Z0 as shown in Figure 1.21. This ratio is the
acoustic impedance of the material carrying the sound wave.
Z0 is constant for a given medium.
Think of it like the gears on a bike:
• The force exerted on the pedal ≡ pressure

• Speed of the pedals ≡ particle velocity
High gear Large pressure and low response to pedal push
Low gear Small pressure gives big response to pedal push
High Z0
p
v
Low Z0
FIGURE 1.21 The relationship between pressure amplitude and the particle velocity for two
materials with a large and small acoustic impedance Z0.
Characteristic Acoustic Impedance, Z0 21
KEY CONCEPT
Acoustic impedance Z0 is important because it is the difference in Z0 between
different materials that determines the size of the echo at an interface.
Table 1.2 shows the value of Z0 for a range of materials.

We cannot measure the excess pressure and the particle velocity of molecules in
a material very easily. In terms of what we can measure, Z0 depends on the stiffness
(bulk modulus) K and density ρ of the medium and can be calculated by measuring
the density and speed of sound, c.
K
Z0 .c K
A dense, rigid material will have a large Z0, whilst a less dense non-rigid material
will have a low Z0.
NOTE
Many texts refer to acoustic impedance as the ‘resistance’ to the sound wave by the
medium. Resistance is usually associated with energy loss, for example, electrical
resistance causing heating of a wire. However, when considering echoes, acoustic
impedance is not associated with any energy loss. Understanding that acoustic imped-
ance is the ‘response of the molecules’ to the sound wave is much to be preferred as
it is then not confused with the energy losses that do occur as sound travels through a
medium. The energy losses are due to attenuation of ultrasound which will be discussed
in Chapter 2 (in electrical terms, acoustic impedance is the analogue of reactance).
TABLE 1.2 Speed of Sound, Density, and Acoustic Impedance Z0 for Several
Tissues and Materials
Material Speed of Sound Density Z0 (kg·m2·s−1 × 106)
(m·s−1) (kg·m−3)
Air 330 1.3 0.0004

Water 1480 998 1.48
Plastic (Perspex) 2730 1180 3.22
PZT transducer 4560 7750 35.34
Fat 1450 911 1.32
Liver 1550 1079 1.67
Kidney 1560 1066 1.66
Blood 1570 1057 1.66
Muscle 1580 1090 1.72
Bone (cortical) 3500 1908 6.68
ENERGY IN A SOUND WAVE
Energy (joule J = kg m2s−2) (recall: energy ∝ p2, where p is acoustic pressure)

Power (watt, W or J·s−1) is the rate at which energy is transferred.
Total transmitted power is the total sound energy emitted by the transducer
summed over all directions.
Recall: The sound source is a piston doing work moving backwards and forwards.
It makes the molecules in front move, and energy is propagated away from the trans-
ducer as a sound wave.
Energy is not generally transmitted equally in all directions. For example, we
want a narrow beam of ultrasound for imaging. We, therefore, need to know how
much energy is being transferred in each direction away from the transducer. For this,
we need to know the intensity of ultrasound.
Intensity (W·m−2 or mW·m−2 or J·s−1·m−2 or kg·s−3) is the power per unit area. It
is the energy flowing across an imaginary surface cutting across the ultrasound beam
as shown in Figure 1.22.
One way of thinking about intensity is to think of how we might distribute a
brush full of paint. The paint on the brush is the total power. We might spread the
paint very thinly over a large area or we might put a thick blob of paint in one spot
and the rest of the area has no paint at all. The thickness of the paint at each point is
the intensity at that point.
Intensity will vary as we measure it at various points in the sound field.
Power = energy per second (radiated by transducer)
Intensity = power crossing 1 cm2 (e.g. at a sample point in the beam)
As the intensity is the energy crossing an area perpendicular to the beam axis, its
value will also be proportional to the acoustic pressure squared.
i.e. I ∝ p02
1 cm
1 cm
Energy flow
FIGURE 1.22 The relationship between power and intensity in an ultrasound beam. Intensity is
the energy flowing across a unit area in the beam.
Decibels 23
NOTE
Intensity is a measure of the power delivered to a point, e.g. in tissue. It is, therefore,
the intensity that has most bearing on safety issues relating to medical ultrasound
(see Chapter 11).
DECIBELS
It is often useful to express the change in energy, intensity, or power from one point
to another as a ratio. As these changes can vary over a very large range of values, for
example, the echo signal from the blood may be 10 000 less than that from a liver–fat
interface; it is easier to use a logarithmic scale to compare values. The decibel scale
(dB) is such a logarithmic measure. It is fully explained in Appendix 2.
A useful way to remember what the ratio is from the value in decibels is
to remember
• A doubling = 3 dB and 10× = 10 dB

• A negative decibel value means a fraction, so −3 dB = a half and −10 dB = a tenth.
CHAPTER 2
The Interaction of Ultrasound
with Tissue
From the ultrasound transducer, pulses of ultrasound are sent into the body and the
sound interacts with the structures within the tissues in a number of ways. We can
look at a single pulse propagating through the body and consider what happens on its
journey. What happens will affect what we see on the image and in this chapter, we
will point out the clinical significance of these factors but will discuss their implica-
tions more fully in Chapter 7 on image quality and artefacts.
REFLECTION AND TRANSMISSION AT A PLANE INTERFACE
In Chapter 1 we defined acoustic impedance Z0 and said that it was differences in

acoustic impedance between two mediums that caused ultrasound to be reflected at
an interface. This phenomenon is key to the formation of ultrasound images as it is
ultrasound reflected back to the transducer that is detected and forms the signal from
which the image is produced.
Specular Reflection
Reflection at a plane interface between two tissues with different acoustic impedances
Z is called specular reflection from the Latin ‘speculum’ for a mirror.
Assuming perpendicular incidence of a plane wave on to a plane smooth inter-
face, some of the energy in the sound wave is reflected and some will be transmitted

25
26 Chapter 2 The Interaction of Ultrasound with Tissue
Medium 1 Z1 Z2 Medium 2
Ii
It
Ir
FIGURE 2.1 The intensity reflected Ir and transmitted It from an incident sound wave of
intensity Ii at an interface between two materials with acoustic impedances Z1 and Z2.
across the interface and go on to deeper structures (Figure 2.1). If medium 1 has
acoustic impedance Z1 and medium 2 has acoustic impedance Z2, then the reflected
intensity Ir as a proportion if the incident sound intensity Ii is given by
2
Z1
2 1
Ir Z2 Z1 Z2
Ii Z2 Z1 Z1
1
Z2
In other words, the ratio of Z2:Z1 determines the echo strength, and that depends
on the density and speed of sound of the two mediums (recall Z = ρ·c).
There is no energy loss at the boundary, so what is not reflected is transmitted into
the second medium.
It Ii Ir
NOTES
• The order of Z1 and Z2 does not matter as the function is squared. If (Z1 − Z2)
is negative, squaring gives a positive value.
• If Z1 = Z2 i.e. if ρ increases and c decreases such that Z stays the same, then
there will be no reflection even though the interface is between two differ-
ent materials.
• If pressures are being measured, rather than intensity, then
Pr Z2 Z1
Pi Z2 Z1
so no squaring – and then the order of Z1 and Z2 does matter, as will be seen
when we look at the construction of a transducer in Chapter 4.
Reflection and Transmission at a Plane Interface 27
Reflection at an Angle and Refraction

If the incident sound wave is not perpendicular to the interface, then the reflected
wave will be reflected at an angle away from the incident direction of propagation.
The transmitted wave will also travel in a different direction from the incident wave
due to a phenomenon called refraction (Figure 2.2).
The angle the reflected wave travels away will be equal to the angle the incident
wave arrives at but on the other side of the perpendicular line. This is expressed in the
Law of reflection – ‘The angle of reflection equals the angle of incidence’.
Reflection can produce artefacts in the image due to the misplacement of reflected
targets in the image.
NOTE
A smooth interface reflecting echoes will only be seen when it is normal to the
direction of propagation and the echoes come directly back to the transducer. A
rough interface will be seen over a wider range of angles as sound is reflected into
a range of angles. However, a rough surface will still be seen most clearly when
perpendicular to the ultrasound beam (Figure 2.3).
Incident Reflected
wave wave
θ1 θ1
Medium 1 Speed of sound C1

Medium 2 Speed of sound C2
θ2
Transmitted
wave
FIGURE 2.2 The angles of reflection θ1 and refraction θ2 of a sound wave incident at a boundary
between two materials with different speeds of sound c2 < c1.
(a) (b) (c)
FIGURE 2.3 Showing reflection of echoes from a smooth interface when the incident wave
is not perpendicular to the surface (a), compared to a perpendicular incidence (b), or a rough
interface (c).
The transmitted sound is refracted. This means that the forward transmitted
beam travels at a different angle to the line of the incident beam. The angle of refrac-
tion depends on the difference in the speed of sound between the two mediums and
is described by Snell’s law.
Snell’s law of refraction shows that the ratio of the sine of the angles of inci-
dence and refraction is given by the ratio of the speed of sound across the interface.
sin 2
c2
sin 1
c1
As shown in Figure 2.4a, when c2 is less than c1, the beam is refracted towards
the normal of the interface and in Figure 2.4b, when c2 is greater than c1, the beam is
refracted away from the normal of the interface.
In order to remember which way the angle changes, think of soldiers marching in
formation on a hard surface moving on to a loose gravel surface (Figure 2.5). As they
march on to the softer gravel surface, they start to move forward more slowly. The line
of direction of the squad moves towards the normal at the change in surface. That is,
crossing on to a slower speed of sound moves the direction of propagation towards
the normal.
Incident wave
θ1 θ1
C1 C1
C2 C2
θ2
C2 < C1 θ2 C2 > C1
Refracted wave
FIGURE 2.4 The angle of refraction at an interface between two materials when speed of sound
(a) c2 < c1 and (b) c2 > c1.
Concrete
gravel
FIGURE 2.5 Illustration of the effect on lines of soldiers when marching from a hard surface on
to the gravel.
Poor Visualisation 29
In both mediums the frequency c of the sound wave is the same, so when the
sound travels into a medium with a slower speed of sound, the wavelength λ of the
sound waves gets shorter.
c
Recall : c f i.e.
f
NOTE
When the angle of incidence is 90° to the interface, the wavelength will still change
as the speed of sound changes and the frequency stays constant, but there will be
no refraction.
REMEMBER
• The reflection coefficient involves acoustic impedance Z1 and Z2.
• The angle of refraction involves the speed of sound c1 and c2.
Refraction can produce distortions in the ultrasound image as the speed of sound
is not exactly the same in all soft tissues (c = 1540 ms−1 ± 5%). For example, the inter-
face between a layer of fat and muscle may be distorted in the image due to the small
difference in the speed of sound.
POOR VISUALISATION
In some patients, complex changes in the speed of sound within overlying tissue can
distort the ultrasound beam to the extent that the image becomes difficult to interpret.
Some types of fatty tissue consist of dense globules of fat in a matrix of less dense
fat. What was a well-behaved narrow beam becomes wider and more poorly defined
due to refraction. This causes poor visualisation of deeper structures in the body. It is
often seen in very obese patients but is not restricted to this body type. It just depends
on the type of fatty tissue. Different positioning of the probe on the skin may help,
but in many cases, it is a physical limitation of the ultrasound technique and cannot
be overcome.
See Chapter 7 on artefacts for illustrations of these effects. Table 2.1 gives exam-
ples of the reflected intensity at various tissue interfaces.
Summary of reflected intensity:

Soft tissues 1% or less (< −20 dB)
Bone – soft tissue 50% (~ −3 dB)
Air – soft tissue 99.9% (0 dB)
TABLE 2.1 Reflected Intensities at Interfaces Between Different Tissues

Interface Reflected Intensity (%) dB
Fat – muscle 1.1 −20

Blood (plasma – RBC) 0.005 −43
Bone – muscle 41 −4
Water – soft tissue (average) 0.23 −26
Air – soft tissue (average) 99.9 −0.004 (≈0)
PZT transducer – soft tissue (average) 80 −1
NOTES
• As almost 100% of the ultrasound is reflected at an air interface, no air can
be allowed between the probe and the skin. Hence, the use of gel on the skin
excludes any air in the gap between the probe and the skin.
• Any gas within the body will block the path of the ultrasound beam, so we
cannot see into air-filled lungs, bowel gas, etc.
• Bone reflects 50% of the ultrasound incident on it. The ultrasound that does
penetrate into the bone is quickly attenuated (see below). Consequently,
we cannot image the inside of the bone at higher frequencies. Some low-
frequency techniques (~1 MHz) where there is some transmission through
bone have been used to examine bone e.g. in osteoporosis. Where bone is
thin access to deeper tissue is sometimes possible, for example, looking
through the temporal bone to image cerebral blood vessels and looking at
soft bone in the foetus.
• The reflectivity of blood is very low −43 dB, so moving blood appears black
on ultrasound images.
SCATTERING
As ultrasound passes through tissue, what is known as Rayleigh scattering occurs

when the sound wave encounters microscopic structures that are much smaller than
a wavelength (e.g. <0.3 mm at 5 MHz).
That is scattering occurs when the target size a .
As a target gets much smaller than the wavelength of the sound wave, sound
energy is not directly reflected but is scattered in all directions. Only a small proportion
of the scattered sound wave will return to the transducer. When the target size is
Scattering 31
Incident
sound
Target << λ Target ≃ λ
FIGURE 2.6 Diagram of examples of Rayleigh scattering from very small targets such as are
seen in tissue parenchyma.
similar to the size of the wavelength, the scattering pattern is complex and depends
on the precise shape and size of the scattering target (Figure 2.6).
Total Scattered Power

Total scattered power is proportional to the target size a (diameter) to the sixth
power and ultrasound frequency f to the fourth power
Scattering a6 f 4 i.e. it is strongly dependent on frequency.
The clinical consequence of this is that frequency is an important factor in the

penetration of ultrasound as it travels through tissue. High frequencies are more
strongly scattered out of the ultrasound beam.
The Speckle Pattern

Ultrasound is a coherent imaging technique. That means that the phase of the
waves can be fully specified at each point in space. In other words, at any one instant
we could calculate where all of the peaks and troughs in the waves are. This may be
compared with normal room lighting where each point in the room receives the light
of many phases at any one time, reflected from many surfaces. Laser light is an optical
example of a coherent light source.
Because ultrasound is a coherent imaging technique, scattering gives rise to the
speckle pattern seen in images. This is the random pattern of dark-light spots at
sub-millimetre level seen within tissue on the image. As shown in Figure 2.7, at the
cellular level, the tissue has many small targets that are smaller than the wavelength
of sound (λ = 0.3 mm at 5 MHz). Each of these targets, that lie within the length of
the ultrasound pulse (e.g. 1.2 mm), will scatter the sound at the same time and send
their tiny echoes back to the transducer simultaneously. These echoes have all trav-
elled back over slightly different distances, so they arrive with slightly different phases
from one another. However, the transducer will see all these echoes at once, and the
signal that is detected at any one instant in time will be the sum of all the peaks and
troughs of the scattered echoes it receives at that instant.
Pulse length
Signal Speckle
pattern
FIGURE 2.7 The origin of the speckle pattern seen in images as sound is scattered from multiple
small targets within the ultrasound pulse.
FIGURE 2.8 Image of the right thyroid showing an enlargement of the speckle pattern.
This adding together of waves is called superposition. One wave is superim-

posed on another. So, at one instant all the peaks add up to give a strong signal, and at
another instant the peaks add to the troughs cancelling out to give no signal. At other
points, the signal strength lies in between. The result on the image are pixels whose
brightness varies in a random way to form a speckle pattern. An example of a speckle
is shown in Figure 2.8.
An example of the superposition of waves is seen when watching sea waves reflect
off a harbour wall. The reflected waves pass through the incoming waves and at one
instant there will be a large peak as the two peaks add up and at another instant the
peak of one will add to the trough of the other and for an instant the sea is flat as they
cancel out.
An optical example of a speckle pattern may be seen by shining a laser pointer at
a very shallow angle on a surface. A speckle pattern is seen because the laser light is
coherent. In this case, it is the microscopic roughness of the surface as it reflects the
light that gives the speckle we see.
Scattering 33
Phase difference
2 0°
1
+ =
2 90°
+ =
+ = 180°
FIGURE 2.9 The superposition of waves with different phases.
At a phase angle of 0°, peaks and troughs exactly add together to give twice the
amplitude. At 180°, they exactly cancel out (Figure 2.9).
NOTES
• Important – We are not seeing the microscopic structure of the tissue in
the image. What we see is a speckle pattern that depends on the frequency
of ultrasound, the pulse shape, and the underlying structure of the tissue.
The tissue structure is too fine to resolve as individual targets.
• Different tissues will alter the appearance of the speckle pattern, so metas-
tases and tumours will look different to normal liver, and the speckle pattern
in the liver will look different to that of the thyroid (Figure 2.10).
FIGURE 2.10 A hepatocellular carcinoma showing a different speckle pattern to that of the
surrounding liver tissue. Source: Minami et al. [1], Baishideng Publishing Group Inc.
• Lower frequency ultrasound will give a coarser speckle pattern due to the
longer wavelength and larger distance needed to change the phase within
each pulse (Figure 2.11).
2.8 MHz LOGIQ

LOGIQ 5 MHz
E9 E9
FIGURE 2.11 The liver was imaged at two frequencies: (a) 2.8 MHz and (b) 5 MHz showing a
coarser speckle pattern at the lower frequency.
ATTENUATION
(Note: Do not confuse attenuation with acoustic impedance.)

This is the overall process by which the intensity of a wave or pulse decreases
with distance from the source. It is the equivalent of the resistance of a medium to the
passage of sound in that it describes the loss of energy as sound travels through the
medium (cf. acoustic impedance in Chapter 1).
In addition to attenuation, there will be boundary losses – i.e. energy reflected
away from the beam at interfaces (specular reflection).
Three mechanisms can be considered under attenuation:
1. Absorption
• Some of the energy in the sound wave causes the molecules in the tissue to
vibrate in a random way that is unrelated to the propagating sound wave.
This is equivalent to heating the tissue. It is non-coherent vibration and is a
loss of energy from the ultrasound beam.
• For low megahertz frequencies, absorption is the predominant attenua-
tion mechanism, accounting for 90% with scattering contributing no more
than 10% [2].
Attenuation 35
2. Scattering
• Energy is re-directed out of the beam by Rayleigh scattering. It is not directly
converted into heat as in absorption although it will eventually be absorbed
by heating tissue.
• At low megahertz frequencies, less than 10% of the acoustic energy is scat-
tered in soft tissue.
3. Beam divergence
• As the beam travels away from the transducer, it will tend to spread out. The
intensity of the beam will decrease as a result. This is not strictly an attenua-
tion mechanism, and it depends on the beam shape (Recall: the definition of
intensity – energy/unit area – as the beam spreads out intensity reduces, e.g.
a pot of paint covers 1 m2 very thickly or 10 m2 very thinly).
NOTE
Absorption and scattering increase with frequency, so high-frequency ultrasound
has poorer penetration than low-frequency ultrasound.
Attenuation Coefficient (α)

(Note: α does not include beam divergence.)
In a uniform medium, due to absorption and scattering, the intensity will decrease
by the same fraction for every centimetre travelled. This gives rise to an exponential
curve (Figure 2.12).
We could say intensity decreases by a factor of (say) 0.5 for every cm – but it soon
gets hard to calculate what the intensity will be at a specific distance from the transducer.
Intensity
1
2
1
4
1
8
Half depth
FIGURE 2.12 The change in intensity with depth due to attenuation in a uniform medium.
Therefore, we use decibels (dB) to express the attenuation coefficient. They

express a ratio in a way that can be added rather than multiplied (see Appendix B).
Say, the attenuation is −3 dB cm−1 (Recall: a ratio of 0.5 = −3 dB)

Then, 8.3 cm of material = 8.3 × −3 dB = −24.9 dB (Easier than trying to do that
using a factor of 0.5/cm)
As well as the distance travelled by the ultrasound, the attenuation coefficient

will also depend on frequency – high frequencies have higher absorption and greater
scattering.
For soft tissue, attenuation is proportional to frequency. Therefore, we use an
attenuation coefficient α that gives the loss of sound wave amplitude in decibels per
cm per MHz. Examples are shown in Table 2.2.
Attenuation coefficient dBcm 1 MHz 1
NOTES
• This is the ‘one-way’ attenuation value, not ‘go and return’ of echoes.
• The attenuation coefficient (α) is the coefficient for change in amplitude of
the sound wave. For a change in energy or intensity, use 2α.
Anisotropy is the phenomena of attenuation (or any other property such as

speckle pattern) varying depending on the orientation of the target to the ultra-
sound beam or image plane. It results from the underlying structure of the tissue,
for example, muscle fibres, having a longitudinal structure along their length that is
TABLE 2.2 The Attenuation Coefficient in Different Tissues

Tissue Attenuation Coefficient, α
Blood 0.2
Fat 0.6
Brain 0.8
Liver 0.6
Kidney 1.0
Muscle (along fibres) 1.3 Anisotropy
(across fibres) 3.3
Water 0.02
Skull bone 20
User Control 37
different to their cross-sectional structure. Their scattering and reflectivity, therefore,

change with their orientation to the ultrasound beam.
Summary
• Attenuation varies with frequency.
• There is considerable variation between tissue types, but the average for soft
tissue is 0.7 dB cm−1 MHz−1.
• The bone is highly attenuating, so 50% of the pulse energy is reflected at the
bone interface. Ultrasound transmitted into the bone is quickly attenuated, and
we do not see inside the bone unless low frequencies are used.
THE JOURNEY OF THE ULTRASOUND PULSE
We can now summarise the journey of an ultrasound pulse as it travels through

tissue (Figure 2.13). Some energy is reflected at large interfaces as specular reflection,
some is scattered by structures smaller than a wavelength of sound, and some will be
absorbed as heat in the tissue. When all the energy in the pulse has been dissipated,
there will be no more echoes returning and the only signal in the receive circuit of
the scanner will be electrical noise. This then defines the useful depth of penetra-
tion for the probe at that transmit power. As high-frequency ultrasound has greater
attenuation, in general, higher frequency transducer probes will have poorer pene-
tration than lower frequency probes. The depth of the target of interest will therefore
influence probe choice.
USER CONTROL
Transmit Power
Transmit power is a control that the user can adjust. We could increase the transmit
power to give greater penetration before the energy was dissipated but that could
have safety implications, for example, physiotherapy uses high-power ultrasound to
Absorption
Specular Scattering
reflection
FIGURE 2.13 Diagram of the ‘ultrasound journey’ through tissue.

actually affect the tissue. For diagnostic imaging, we must limit the power output of
the transducer, which in turn limits the depth of penetration.
The reduction in signal strength with depth, due to attenuation, can be compen-
sated for by increasing the amplification or gain applied to the echo signal. This is
fully discussed in Chapter 5.
Through its interaction with tissue, ultrasound is able to probe and give
information on many physical properties of tissue, such as its speed of sound, stiff-
ness, and attenuation. These in turn provide diagnostic windows into the tissues,
shedding light, and giving information on the physiological and pathological status
of the tissue. Quantitative measurements may be obtained and changes can be moni-
tored. Later chapters will look at some of these possibilities in more detail [3].
REFERENCES
1. Minami, Y. and Kudo, M. (2010). Hepatic malignancies: correlation between sonographic
findings and pathological features. World Journal of Radiology 28: 249–256.
2. Duck, F.A. (1990). Physical Properties of Tissue: A Comprehensive Reference Book, 73–135. London,
UK: Academic Press.
3. Cloutier, G., Destrempes, F., Yu, F., and Tang, A. (2021). Quantitative ultrasound imaging of soft
biological tissues: a primer for radiologists and medical physicists. Insights Imaging 12: 127–147.
https://doi.org/10.1186/s13244-021-01071-w.
CHAPTER 3
Beam Shapes
In this chapter, we look at the beam shapes used for clinical ultrasound imaging. In
Chapter 4, we will look at the transducers used to produce these beams.
The ultrasound beam can be thought of as the ‘corridor’ along which ultrasound
(continuous transmission or pulses) is sent out and along which the echoes travel
back to the transducer. The following discussion assumes continuous wave (cw)
transmission at a single frequency. Pulses will be considered later.
If the source of a sound is a point source – by which we mean it is much smaller
than the wavelength of the sound – then the sound will spread out in all directions
(Figure 3.1). An example of a point source is a stone thrown into a pond. It sends rip-
ples equally in all directions, the stone being smaller than the wavelength of the rip-
ples. Another example is when you speak. The wavelength of sound in air is ~0.5 m.
As this is larger than the size of your mouth, the sound spreads in all directions from
your mouth and someone standing behind you can hear you.
If the source of a sound is an infinite plate moving backwards and forwards (cf.
the piston), then the sound would propagate away from that plate only in the forward
direction as plane parallel waves.
The ultrasound transducers we use fall somewhere in between these two extremes.
We want narrow beams of ultrasound so that when we point our ultrasound beam at
a particular target, we only get echoes from that target and nowhere else. The width
of the radiating surface of the transducer is called the aperture. Note that for ultra-
sound transducer arrays, the aperture at any one time may be less than the full width
of the transducer, that is, only part of the array is being used.

39
40 Chapter 3 Beam Shapes
Point source Plane waves
FIGURE 3.1 Wavefronts generated by a point source and an infinite plane source.
SIMPLE BEAM SHAPE MODEL
To a first approximation the shape of an unfocused ultrasound beam, produced by a

plane (flat) transducer, consists of two distinct zones, as shown in Figure 3.2.
The near field or Fresnel zone has a uniform diameter along its length. In the
far field or Fraunhofer zone, the beam spreads out in a uniform manner.
a2
The length of the near field is D ,
where a is the half aperture width and λ is the wavelength of sound.

The spread of the beam in the far field is
sin for a square transducer

a
or
sin 0.6 for a circular transducer

a
where θ is the half angle of the spread.
Near field Far field

(Fresnel zone) (Fraunhofer zone)
a θ
FIGURE 3.2 Diagram of the simple model of an ultrasound beam.

Simple Beam Shape Model 41
TABLE 3.1 Examples of Transducer Parameters for a Plane Square Transducer

λ (mm) f (MHz) a (mm) D (MM) θ
0.3 5 10 333 1.7°

0.3 5 5 83 3.4°
0.3 5 1 3 17.5°
1.5 1 10 66 8.6°
1.5 1 5 16 17.5°
1.5 1 1 0.6 Not defined
Table 3.1 shows what this means for a plane square transducer using the follow-
ing values:
c = 1540 ms−1 (a) f = 5 MHz λ = 0.3 mm

(b) f = 1 MHz λ = 1.5 mm
NOTE
sin(90°) = 1; so, if the aperture 2λ, then λ/a > 1. As sin θ cannot be greater than 1,
the source will act as a point source and sound will radiate in all directions.
SUMMARY
• A large aperture for a given wavelength λ gives a uniform beam with a long
near field and a small spread in the far field.
• As the aperture approaches the size of one wavelength, the source becomes
like a point source with a short near field and a widely spreading far field.
• Narrow beams require a wide aperture compared with the wavelength.
• We can reduce the aperture size and still have a narrow beam by reducing
the wavelength, i.e. we increase the frequency.
In order to have narrow beams, imaging ultrasound uses high-frequency ultra-

sound for a given aperture.
NOTE
Most ultrasound imaging uses wide enough apertures so that it uses the near field
for imaging where the beam does not diverge over the usable depth of penetration.
In this region, the beam can also be focused (see below).
The ‘near field’ – ‘far field’ picture is an approximation of the beam shape. If we
plot the intensity within a real ultrasound beam from a plane (flat) transducer, we
(a) First side lobe
Complex pattern Main lobe

of peaks + troughs
Intensity contours
(b)
1
Decrease
r2
r
(c)
θ
a
∼25% –10d
narrowing B co
ntou
r
Last axial
maximum
FIGURE 3.3 (a) Intensity contour map of a typical continuous wave ultrasound beam from
a plane square transducer, (b) the axial intensity, and (c) comparison with the simple beam
shape model.
get the diagram shown in Figure 3.3. The lines are lines of constant intensity, that is,
it is an intensity contour map of the beam. The centre plot (b) is the axial inten-
sity – along the central axis of the beam. The lower diagram (c) compares these two
with the simple diagram of near field and far field we have already considered.
In the plot of the intensity changes within a real beam, we see a complex pattern
of changes in the near field with a single maximum peak on the axis at the transition
from the near to the far field. At this point the beam narrows by about 25%, equivalent
to a natural focusing of the beam. Beyond this point, in the far field, the beam spreads
out. We also see side lobes spreading out from the main beam just beyond the last
axial maximum (near field). Echoes arising from the side lobes will also contribute to
the received signal. As they are not from the main beam their signal produces clutter
in the image, degrading the diagnostic image data. Figure 3.4 shows a beam plot from
a 4 MHz plain transducer.
Huygen’s Wavelet Model and Diffraction 43
Sound pressure in water (4 MHz, D = 10 mm)

10 0 dB
5
–5 dB
x(mm)
–10 dB
–5
–10 –15 dB
20 40 60 80 100 120 140 160 180 200
z(mm)
FIGURE 3.4 Beam plot in water of an unfocused ultrasound beam from a 10 mm diameter
4 MHz plane transducer.
NOTE
The intensity variations are less pronounced for short pulses than for cw ultra-
sound because short pulses contain a range of frequencies and therefore tend to
smooth out the fine variations in intensity.
The concept of a beam applies to both transmission and reception. In other

words, received echoes travel back to the transducer along a beam with similar
properties to the transmitted beam we are considering. Transmission and reception
are therefore reciprocal. However, in a real scanner, the receive beam does not have
to be the same as the transmitted beam. We can set up a different beam shape for
reception than that used for transmission (see Chapter 6). The net beam shape for
the transmit-echo round trip will then be the transmit and receive beams shapes
multiplied together.
HUYGEN’S WAVELET MODEL AND DIFFRACTION
A useful way of understanding how the beam shapes we see arise is to use Huygen’s
wavelet model (Figure 3.5). A plane transducer aperture can be modelled as many
point sources next to each other. These are known as Huygen’s sources and their
waves as Huygen’s wavelets. The Huygen’s wavelets from each point source spread
out spherically from that source (Figure 3.5a). As shown for the transducer with a
limited aperture (Figure 3.5b), the wavelets from Huygen’s sources across the aper-
ture add by superposition so as to give the expected plane wave in front of the ‘piston’,
but at the edges they spread out sideways with no more sources to interact with. The
pattern so produced is called the diffraction pattern for that source or aperture.
The interaction of the wavelets with one another, through superposition, is known as
interference. In general, the shape of the beam we get from a transducer (if there is
no reflection, refraction or attenuation) is the result of the diffraction pattern from the
source or aperture of the transducer.
Hygens source
(point source)
Constructive
Hygens wavelets
interference
FIGURE 3.5 Huygen’s wavelets for (a) a point source and (b) a plane transducer aperture.
Where a peak adds to a peak to give larger intensity, it is known as constructive

interference and where a peak and trough cancel each other out, it is known as
destructive interference.
Using Huygen’s wavelet model enables us to see how the various beam shapes
used in ultrasound arise. A more detailed explanation of how the shape of the beam
we see arises is given in Appendix C.
FOCUSING
We have seen that a plane unfocused transducer has some narrowing of the beam at
the last axial maximum of ~25%. This is a natural focusing.
Consider the idealised geometrical arrangement of a spherical bowl-shaped trans-
ducer shown in Figure 3.6.
Huygen’s wavelets generated from across the transducer aperture constructively
interfere to produce converging wavefronts that all arrive at the same time at F. So, F
sees a high intensity just at that point. At other field points in that range, there will be
destructive interference and low intensity or no sound wave at all.
In a real physical situation, because of diffraction at the edges of the aperture, a
spherically curved transducer will produce a narrow beam at a focal point, but the
beam will still have a finite width as shown in Figure 3.7. The use of an acoustic lens,
made of plastic or rubber, similar to a glass lens in optics, can also be used to focus an
ultrasound beam.
The width of the beam at the focus is W F ,

a
where F is the focal length, a is the half aperture width, and λ is the wavelength
of sound. Figure 3.8 shows a focused beam from a linear array.
Focusing 45
F F
(a) (b)
FIGURE 3.6 Idealised geometrical focus from a spherical transducer shown as (a) geometrical
rays and (b) converging wavefronts.
Focal zone
FIGURE 3.7 A spherical transducer showing the definition of the focal zone and focal width W
for a focal depth F.
NOTES
• The focal zone or axial length of the focus is defined as the region where
the beamwidth is 2W.
• For a focused transducer, the focal point equates to the far field of an unfo-
cused transducer.
• A large aperture gives a narrow focal width just as a large aperture gives a
small beam spread (sin θ) for an unfocused far field.
• The focal length (distance from the transducer to the focus) must be less
than the near field length of the unfocused transducer. In other words, you
can only focus the beam in the near field. By using wide aperture trans-
ducers, the near field will be longer and focusing can be used throughout
the useful range of the image. Ultrasound beams from clinical imaging
probes generally use the near field region for imaging (Figure 3.8).
MM
FIGURE 3.8 Beam plot of a Doppler pulse from a linear array with a 2.9-cm focal depth. The
first centimetre nearest the transducer is not shown. Source: Courtesy of Barry Ward.
Strong focusing
Weak focusing
No focusing
(plane transducer)
D D
2 2
FIGURE 3.9 The difference in beam shape with strong and weak focusing.
The strength of focus may be defined as the ratio of the aperture width a to the
width of the beam at the focus W.
a a2 Dunfocused
where D is the near field length.
W F F
Where the focal length is less than half D we have strong focusing, and where it is
greater than half D we have weak focusing (Figure 3.9).
NOTES
• For weak focusing, the focal zone is longer than that for strong focusing.
• There is an improvement in beamwidth at the focal zone but beyond that
the beam spreads out more than it would for an unfocused beam. This has
implications for visualising tissue in front of or beyond the focal zone. The
target of interest should be in the focal zone (see below).
• To have strong focusing at depth, you need a wide aperture.
BEAM FORMING WITH TRANSDUCER ARRAYS
By using arrays of elements to build a transducer, all aspects of beamforming can be

controlled electronically. This includes controlling the aperture width and position,
steering and focusing. These can all be changed in real time, that is, as the image
is acquired.
As we consider simple arrays, it is important to note that the ultrasound beam
will have a beam shape in the scan plane or image plane, that is determined by how
the elements of the array are driven, and a beam shape in the elevation plane that
will be the beam shape of a single transducer element (Figure 3.10). It is this elevation
plane beam shape that will determine the slice thickness of our image. It is impor-
tant to recognise that the image produced is not an infinitely thin cut through the
tissue. The slice thickness can be optimised by using an acoustic lens in the front of
the transducer so as to give weak focusing throughout the elevation plane.
It is easiest to begin by considering a linear array that will produce a linear scan
image, as we saw in the introduction (Figure 0.1a).
Figure 3.11 shows a small array of four elements.
Elevation
plane
Cylindrical
Scan plane
lens
FIGURE 3.10 Definition of scan plane and elevation plane for a linear array transducer.
4-Element array
FIGURE 3.11 A 4-element array with beams from each element indicated.
NOTES
• Each element in the array has its own beam shape, which is the beam shape
produced by a single element transducer as discussed above.
• The array elements are small so the beam shape from each element is wide
and spreads over a wide far field angle.
• Wavefronts from all the elements combine by interference to give the
wide aperture beam shape we want to make from the array as shown in
Figure 3.12.
FIGURE 3.12 Wavelets from a 4-element array showing the alignment of wavefronts to give
plane waves from the whole array.
The strongest reinforcement of intensity is the tangent of all the wavefronts, i.e.
straight ahead. This looks like the beam shape we would get from one large aper-
ture. However, in an array, although each element will have its own beam shape
that approximates a point source, there are no waves being generated from the gaps
between the elements.
Grating Lobes
Analogously to the formation of side lobes (see Appendix C), an array can produce
grating lobes due to diffraction effects between the individual elements. A grating
lobe is a directional lobe of increased intensity where there is positive reinforcement
of waves from individual elements in the array.
Grating lobes arise because the array uses independent elements with gaps bet-
ween them (Figure 3.13). Along the line of the first grating lobe, the wavelets from
each of the elements are in phase as the path difference is exactly one wavelength.
They, therefore, reinforce each other and we have a grating lobe (a secondary beam)
formed away from the main beam axis. In principle, there could also be a second
grating lobe when the path difference is exactly two wavelengths. The angle of the
Main beam Grating lobe
λ λ λ
FIGURE 3.13 Diagram showing the origin of a grating lobe at an angle giving one wavelength
difference in path length.
grating lobe depends on the spacing of the transducer elements. The wider apart the
elements, the closer the grating lobe is to the main beam. The intensity of the grating
lobe is lower than the intensity of the main beam.
The angle of the grating lobe from the main lobe is given by
sin where d is the distance between elements.

d
If the distance between the elements is equal to or less than a wavelength, i.e. if
d λ, then sin−1(θ) is undefined as the value is >1 and sin(90°) = 1. Since an angle of
90° would be along the line of the array, there will be no grating lobe for the beam. In
other words, the elements are then close enough together to make the array behave
like a single transducer. In general, linear and curvilinear arrays used for medical
imaging will have element spacing >λ, so there will be grating lobes (Figure 3.14).
Main beam
G
ra
t
in
g
lo
eb
FIGURE 3.14 The overall beam shape for an array with grating lobes.
Gaussian beam profile
FIGURE 3.15 Using apodization, the beam can be given a Gaussian profile which remains
constant throughout the length of the beam.
The exception is for phased arrays which have fine element spacing of <1λ (see
Chapter 5).
Apodization
For each individual element in the array, the phase and amplitude of the signal driving
that element during transmit can be controlled.
Shaping the beam across the active aperture by tailoring the amplitude to each
element is called apodization. By a careful choice of amplitudes, the sidelobes in
the beam can be greatly reduced. In particular, a beam cross-section profile with a
Gaussian shape has reduced side lobes and maintains its shape along the length
of the beam (Figure 3.15). The beamwidth is slightly wider than for a non-apodized
beam, but the benefits of a more uniform beam at all depths are useful and some scan-
ners use Gaussian beams.
BEAM STEERING
It is useful to be able to steer a beam electronically using an array. This can be

done using time delays to alter the phase and hence the position of the wavefronts
(Figure 3.16).
For example, by delaying the phase of the wavefront by Δt between adjacent
elements, we can cause the wavefronts to ‘add up’ along a direction at an angle to
the straight ahead beam. When the peak transmitted from the first element has
moved out by s in the direction we want to send the beam; the peak from the next
element which has been delayed by Δt is just leaving its element, and the two
peaks are in phase and constructively interfere along the direction we want to
steer the beam.
s c t Recall : velocity distance / time

Beam Steering 51
2Δt
S
Δt θ
θ
d
FIGURE 3.16 Diagram of a steered transmitted beam using time delays to delay the phase of
wavefronts from each element.
So,
t
sin c
d
where d is the distance between elements, and c is the speed of sound in tissue.
Therefore, to steer the beam by 30° (sin (30°) = 0.5) on an array with element
spacing d = 0.5 mm, c = 1540 ms−1
sin
t d so, the phase delay between elements needs to be t 0.11 s.
c
Using the same time delays for the returning echoes, the wavefronts from the
steered direction will be brought back together and be in phase with each other.
Returning echoes reaching each element are delayed by Δt and then summed so
phases of echoes arriving from the steered direction sum constructively. Those from
other directions sum destructively and are not detected (Figure 3.17). The trans-
ducer will then have greatest receive sensitivity to echoes arising from the steered
direction. This technique is called delay and sum beamforming.
Returning
echoes
2Δt
S
Σ Δt θ
θ
d
FIGURE 3.17 Diagram showing how a received steered beam is detected with time delays and
summation of signals from each element.
Beam Shape and Beam Steering
NOTES
• The main lobe has the greatest intensity/sensitivity when it is aligned with
the main axis of the array; in other words, when it is steered forward. This
has an important implication when looking for weak targets with Doppler
and colour Doppler – the greatest sensitivity is obtained when the beam is
steered forward.
• As the beam is steered away from the main axis, the main beam gets weaker
and the grating lobe becomes stronger as it approaches the forward position
(Figure 3.18).
• Phased arrays (see Chapter 6) scan the image plane by steering the beam
through a wide angle. In order to completely suppress the grating lobes,
their elements are spaced <λ/2 apart so that even with steering up to 90°,
sin−1(θ) is >90° for the grating lobe; in other words, it does not exist.
Beam steering also has another effect on image quality. As the beam is steered, the
effective aperture of the transducer is reduced, so the beam will spread more and res-
olution will be reduced (Figure 3.19) (see below for more on resolution).
Figure 3.20 shows a beam plot of a steered beam from a linear array.
ELECTRONIC FOCUSING
A physical lens placed in front of a light source or a sound source causes delays in
the wavefronts so that they bend and converge to form a focus. In a similar manner
to beam steering, the same effect may be produced using time delays between the
Main beam steered
Main beam
forward streered
Grating lobe
Grating lobe streered
FIGURE 3.18 Diagram showing how steering the beam decreases main beam sensitivity and
enhances the grating lobe signal.
Electronic Focusing 53
Ws
FIGURE 3.19 The effect of beam steering on the effective aperture of the transducer.
MM
FIGURE 3.20 Beam plot of a steered Doppler pulse from a linear array with a 5.6 cm focal depth.
Note the side lobes appearing towards the straight ahead direction. Source: Courtesy of Barry Ward.
elements of a transducer array. The timings between the elements are chosen so that
the transmitted wavefront from adjacent elements reinforces along a curved line
which converges on a field point, equivalent to focusing on that point (Figure 3.21).
The effect is as though the beam had been produced by a spherically curved trans-
ducer (see Figure 3.6).
As for steering, by delaying the wavefronts arriving at the transducer with the
same time delays as were used in transmit, the transducer will have the greatest sen-
sitivity for echoes arising from the focal point. After passing through the time delays,
the wavefronts are summed. The delays bring the phase from each element together,
so they reinforce each other for that focal point.
By changing the time delays, the depth of the focus can be controlled. By chang-
ing the width of the aperture (number of elements used), the focal width can be
controlled.
Δt1 Δt1
Δt2 Σ Δt2
Δt1 Δt1
Returning
echoes
(a) (b)
FIGURE 3.21 Diagram showing the use of time delays to produce the converging wavefronts of a
focused transducer (a) in transmitting and (b) in receiving.
RESOLUTION
Ideally, we would like to image organs with very fine detail shown in the image so we
could detect very small but clinically significant changes. However, there is a limit on
how small a target we can clearly image. The resolution of an imaging modality is
the minimum size of the target that can be imaged.
As a pulse travels down the ultrasound beam, it typically has a teardrop shape in
space, shaped by the image plane (x-axis) and elevation plane ( y-axis) beam shapes,
and the pulse length (Figure 3.22).
DEFINITION
Resolution is the ability to resolve two targets as separate targets – rather than
one blob on the image, as shown in Figure 3.23.
C
y
nλ
FIGURE 3.22 The ‘teardrop’ shape of an ultrasound pulse in space.
Target Just resolved Not resolved
FIGURE 3.23 For two small targets closely spaced (a), they are just resolved in the image if they
are seen as two targets (b) and not resolved if they appear as a single target (c).
Resolution 55
Lateral direction (x)
C
A Beam width Axial direction (z)
B
Transducer Pulse length
FIGURE 3.24 Location of points (A, B, C) within the pulse used to define image resolution.
Echo
Z
A A B
FIGURE 3.25 Resolving points A and B along the beam axis.
We need to consider the resolution along the axis of the ultrasound beam, axial
resolution, and in the transverse image plane, lateral resolution (Figure 3.24).
Axial Resolution
Resolution along the direction of propagation A-B (Figure 3.25).
A and B are two locations in the tissue that are both inside the pulse at the same
time, therefore both send back echoes simultaneously. As the pulse passes A in the Z
(time) direction, the echo from A maps out the pulse length shape on the image, i.e.
not a single point but a blob. Similarly for B.
A and B will just be separated if B is just coming into the pulse as A is just leaving
the pulse. This will be the case if A and B are half a pulse length apart, so
n
Axial resolution, A B where n is the number of wavelengths in the pulse.
2
Any targets closer than this will not be resolved as separate targets.
To improve the axial resolution, we would need
• Fewer cycles = shorter pulse
or
• A higher frequency = shorter wavelength, λ
Lateral Resolution
Resolution across the beam in the image planes B-C (Figure 3.26).
B and C are two locations across the beamwidth that are both inside the pulse at
the same time. The ability to separate B and C depends on how wide the beam is, for
x
B C B C
FIGURE 3.26 Resolving points B and C across the beamwidth.
example, at the focus the width is small. As the ultrasound beam is swept across B and
C to form the image, the echo from B maps out the beamwidth profile on the image
producing a ‘blob’, not a single point. Similarly for point C.
As for axial resolution, B and C will just be separated if the distance B-C is half the
beamwidth. In other words, B just leaving the beam as C just enters it.
Any targets closer than this will not be resolved as separate targets.
To improve the lateral resolution, we would need a narrower beam. We could
do this by:
• Using a higher frequency

• Recall: for a plane transducer the near field distance D = a2/λ = a2f/c, so a
higher frequency means we can use a narrower aperture a with a narrower
beamwidth.
or
• Stronger focusing
In order to look at the resolution, we have considered three targets A, B, and C

that the pulse shape sample volume passes. Another way of looking at it is to see that
everything inside the sample volume of the pulse adds together to give the echo we
detect and so all the information from within the sample volume of tissue is super-
imposed in one echo. In other words, that information is degraded. It is these echoes
arising from within the sample volume, i.e. pulse length/2 or beamwidth/2 that give
rise to the unresolved speckle pattern in the image due to Rayleigh scattering (see
Chapter 2). An example of the high resolution that can be achieved using very high
frequencies is seen in Figure 3.27.
NOTE
Lateral resolution will be poorer if line density is insufficient to cover the target (and
potential targets will be missed). For sector shape imaging line density increases
with distance (Figure 3.28). Line density will be discussed further in Chapter 7.
Resolution 57
1 mm
T N
FIGURE 3.27 Anterior chamber of the eye with a prosthetic intraocular lens. An example
of very high-frequency imaging at 50 MHz showing the high resolution of <60 μm that can be
achieved with very high frequencies [1].
FIGURE 3.28 Diagram showing how line density changes with depth for a sector scan.
SUMMARY
In order to have both good axial resolution and good lateral resolution, we need
to use high frequencies. High-frequency ultrasound probes will always give better
resolution than lower frequency probes.
Resolution Versus Penetration

We have seen that in order to have a fine resolution, a high frequency is needed. How-
ever, in Chapter 2, we saw that higher frequencies suffer from higher attenuation due
to greater scattering and absorption and so have less penetration. There is therefore a
trade-off between resolution and penetration.
NOTES
• Use the highest frequency probe that allows visualisation at the depth of the
target of interest.
• Some scanners have a control that lets the user trade between resolution
and penetration. This control alters the frequency of the transmitted pulse
within the bandwidth of the transducer.
CLUTTER
The presence of echoes arising from side lobes, grating lobes, and multiple reflections
from within the body all contribute to the signal that forms the image. Such echoes
do not arrive at the transducer from the main beam and the information from them
is spurious. The term used to describe the degradation of the image caused by these
echoes is ‘clutter’. A number of techniques, such as using Gaussian beams, have been
used to reduce the effect of clutter in the image.
REFERENCE
1. Coleman, D.J. and Silverman, R.H. (2006). Explaining the current role of high frequency
ultrasound in ophthalmic diagnosis. Expert Review of Ophthalmology 1: 63–76.
CHAPTER 4
The Ultrasound Probe
Within medical ultrasound, the terms ‘ultrasound probe’ and ‘ultrasound

transducer’ are often used interchangeably. Strictly speaking, the transducer is the
device within the probe that produces the ultrasound and the probe is the whole
assembly that the user holds and manipulates to produce the images they want.
A transducer is any device that converts energy from one form to another. As
seen in the first chapter, a sound wave is produced if we have a mechanical vibrator
such as a piston or a loudspeaker. A loudspeaker is a transducer converting electrical
energy into mechanical energy – the sound wave.
In order to use ultrasound to produce medical images, we need to convert an
electrical signal into an ultrasonic sound wave and then to convert mechanical
movement produced by the returning echoes into an electrical signal that can be pro-
cessed to form the image.
Figure 4.1 shows the basic layout of an ultrasound probe as used in medical ultrasound.
We look at each of the components in turn.
THE TRANSDUCER
Most medical ultrasound transducers use the piezoelectric effect to convert electrical
energy to sound energy and then sound energy from echoes back to electrical energy.
Piezoelectric Effect
Some materials show the property that when a voltage is applied across them, they
either shrink or expand depending on the polarity of the voltage (Figure 4.2). The

59
60 Chapter 4 The Ultrasound Probe
Matching
Electrodes layer
Shielded
cable
Lens
Sound absorbing PZT transducer

backing layer
FIGURE 4.1 The basic elements of an ultrasound transducer.
–
0 +
–
0
+
FIGURE 4.2 The voltages produced when a piezoelectric material expands and contracts.
phenomenon is called the piezoelectric effect. It also works the other way around.
If the material is squeezed or caused to expand, a voltage forms across the opposite
surfaces of the sample. The voltage is proportional to the degree of deformation of the
material. Quartz is an example of a naturally occurring piezoelectric material. This
forms the basis of the timer in a quartz watch, when a piece of quartz is made to reso-
nate at a high frequency. This produces an alternating voltage that regulates the clock,
as a pendulum does in a grandfather clock.
The piezoelectric effect is ideal for making ultrasound transducers as piezoelec-
tric materials can be made to resonate at megahertz ultrasonic frequencies. They can
be driven by applied voltages to transmit ultrasound (act as a piston on the surround-
ing medium) and when vibrated by the returning echoes, they produce an alternating
voltage that forms the received signal for processing to form an image.
Most modern medical ultrasound transducers use a manufactured ceramic called lead
zirconate titanate (or PZT), which can be formed to any shape required. Another piezo-
electric material sometimes used is PVDF (polyvinylidene fluoride), which is a polymer.
Manufacture of Piezoelectric PZT Transducers
At a molecular level, untreated PZT has a domain structure in which each domain
has a positive end and a negative end. These are randomly oriented. If a voltage
is applied, the domains line up and the PZT sample expands by electrostriction,
The Transducer 61
V– V+
V+ V–
FIGURE 4.3 The behaviour of unpolarised PZT when a voltage is applied before it is made
piezoelectric. Unpolarised PZT (a) expands by electrostriction as domains line up (b).
V+ is attracted
V+ is repelled
kV
+ Polarized V–
V–
V–
–
V+
Applied voltage
FIGURE 4.4 PZT is polarised by cooling under an applied voltage (a). It then exhibits the
piezoelectric effect (b).
whichever way the voltage polarity is applied as shown in Figure 4.3.

In order to make it piezoelectric a high voltage is applied whilst the sample is
heated above a critical temperature called the Curie temperature. The domains
line up under the applied voltage (Figure 4.4). The sample is then cooled whilst the
voltage is maintained and the polarisation of the domains is frozen into the sample.
The sample is then piezoelectric and further application of voltages will produce
the expansion and shrinking seen in piezoelectric materials.
Single Crystal Technology

Fabricating a conventional PZT transducer using the method described in the box
results in a transducer in which, typically, only 70% the material is polarised. More
recently, a significant improvement in transducer efficiency has been obtained by
growing a single crystal of the piezoelectric material. This produces a transducer with
fewer defects and lower losses which when polarised produces a near perfect align-
ment of poles. Effectively, the whole transducer has a single domain for polarisation.
λ
2
FIGURE 4.5 Half wave thickness of a transducer.
This improves efficiency by 85% over conventional PZT transducers allowing greater
sensitivity and improved visualisation in difficult scanning situations.
Resonant Frequency
A slab of PZT will vibrate at its resonant frequency when ‘hit’ with a voltage spike, like
pinging an empty wine glass to make it ring as shown in Figure 4.5. The resonant
frequency will be that for which the slab is a half wavelength thick.
For example, we need a resonant frequency of 5 MHz
Speed of sound in PZT, c = 3200 m·s−1
c 3200
So, 0.64 mm so, slab thickness, / 2 0.32 mm
f 5 106
The resonant frequency chosen will be the centre frequency for the probe.
Thin metal electrodes are deposited on each side of the PZT slab to apply the
transmit voltages and pick up the receive signal. The signal is fed through a shielded
cable to avoid interference from other signals.
BACKING LAYER
We have previously seen that in order to have good resolution we need very short
pulses. Therefore, we need to stop the transducer from ringing when the voltage
spike is applied like pinching the edge of a wine glass whilst pinging it with the other
hand. This is called damping (Figure 4.6). Within the ultrasound probe, damping is
achieved by putting a dense sound absorbing material against the transducer forming
a backing layer in the probe. It is the effect of the backing layer that enables pulses to
be short and the transducer wideband in its frequency range.
NOTE
Because the backing absorbs some of the sound energy and some sound is reflected
back into the transducer at the air–skin interface, the transducer can warm up.
This is known as self-heating and is a potential hazard when placing the trans-
ducer on the skin.
Matching Layer 63
Amp
Vapp t t
Resonant ringing Damped pulse
FIGURE 4.6 The process of impulsing a transducer element with a voltage spike Vapp, showing
resonant and damped ringing.
Recent improvements in the technology and design of backing layers have

improved transducer efficiency and reduced the self-heating problem.
MATCHING LAYER
PZT is a hard, rigid ceramic material with a high impedance and skin is a relatively
low impedance material. Recalling Chapter 2, there will, therefore, be a large reflec-
tion of ultrasound at the interface between the transducer and the body, so potentially
very little sound enters the body. This problem can be overcome to a large extent by
the use of a matching layer. This is a layer of a material whose impedance lies some-
where in between the impedance of the transducer and that of the body. It is the same
feature as the anti-reflective bloom seen on the front of a camera lens or binoculars.
The coloured bloom is a matching layer that allows the light to pass in or out of the
lens without being reflected back into the lens at the glass–air interface.
How Does the Matching Layer Work?
Recall: Reflected intensity
2
Ir Z2 Z1
Ii Z2 Z1
The amplitude is proportional to the square root of intensity, i.e.
Ar Z2 Z1
Ai Z2 Z1
Looking at the equation for amplitude we see that the order of Z2 and Z1 mat-
ters as there is a minus sign there. If we go from a low impedance to a high
impedance, the result is positive and if we go from a high impedance to a low
impedance the result is negative. What this means in terms of the amplitude of
(continued )
(continued)
the reflected wave is that where the result is negative, there will be a phase inver-
sion of the sound wave. What was positive going becomes negative going, i.e. 180°
phase shift. This will happen going into the skin as the skin has a low impedance
compared with the PZT.
The matching layer is made of a material whose impedance lies between that
of the PZT and the body and is a quarter of a wavelength thick. It is therefore also
known as quarter wave matching or impedance matching.
The way the matching layer works is shown in Figure 4.7. Any reflected
sound at the body interface will have its phase inverted, i.e. a positive peak
becomes a trough = λ/2 change. The sound wave then travels back through
the quarter-wave layer = λ/4 change in phase due to a quarter of a wavelength
in distance. It then reflects at the higher impedance of the PZT, so no phase
change, and then travels back across the quarter-wave layer to the body to give
another λ/4 change in phase.
So, the total body–PZT–body reflection path = λ/2 + λ/4 + λ/4 = 1λ
This means that the second, reflected, sound wave T2 transmitted into the body
is in-phase with the first transmitted wavefront T1 and reinforces it. All multiple
reflections will be similarly in-phase and at that one frequency there will be almost
perfect transmission of the sound into the body.
The matching layer is made of a material whose impedance is the geometric
mean of the impedances on either side, i.e.
Z ML Z body Z PZT
Matching
layer
PZT Body
ZPZT ZML ZB
T1 T1
Inverting
reflection
T2
T2
Non-inverting
reflection
T3
λ
4
(a) (b)
FIGURE 4.7 The phase changes occurring in a quarter wave matching layer.
Wide Band Transducers 65
NOTES
1. Quarter wave matching will only work really well for one frequency
f ML
cML
2. Multiple Matching Layers: With each layer having an impedance of

the geometric mean of those on either side, will match over a wider range
of frequencies such as are contained in a short pulse of sound. The effect
is then closer to a gradual transition from ZPZT to Zbody. This is then like
multi-coated lenses on a camera or the bell on a trumpet matching a small
tube to the open air. There is a more gradual transition from a high to a
low impedance.
FRONT FACE LENS
On the front face of the transducer probe is an acoustic lens made of a soft rubber
with an impedance close to that of soft tissue. In the case of a simple array probe,
the lens will be a cylindrical lens across the front face of the probe giving weak
focusing in the elevation plane, to minimise slice thickness, but having no effect in
the scan plane. In addition to giving some focusing in the elevation plane, the front
face forms a hygienic, insulating barrier to the probe, so it can be safely cleaned (see
Figure 4.12).
WIDEBAND TRANSDUCERS
Advantages:
• Very short pulses are possible, giving good resolution.

• It is possible to drive the transducer with complex or ‘special’ pulse shapes.
• Harmonic imaging using higher frequencies is possible.
• The same transducer can use different frequencies for imaging and Doppler.
To make a transducer wideband requires the following:
• Transducer element thickness chosen to be at the middle of the wideband range.

• Efficient damping in the backing layer.
• Very good matching to body impedance.
CONSTRUCTION OF AN ARRAY
Most ultrasound probes used in medical imaging consist of an array of transducer ele-
ments aligned next to each other (Figure 4.8). They may be fabricated from a single slab
of PZT etched to produce the individual elements, but the principle components of
their construction are the same as for a single element transducer, as described above.
NOTES
• For a phased array, the element spacing is <λ/2 so that there are no grating
lobes formed when the beam is steered through large angles.
• The array may be ‘bent’ to form a curvilinear array.
• The development of the technology used in constructing ultrasound probes
has used new materials and techniques to improve their weight, sensitivity,
and resolution. For example, each element may be further split, a method
called sub-dicing. The gaps produced are filled with a low impedance
material thereby reducing the effective impedance of the PZT element. This
improves signal transmission and reception into the body.
CMUT TECHNOLOGY
More recently, it has become possible to construct transducers that operate using a
varying capacitance between electrodes rather than using the piezoelectric effect of
a solid material [1]. The device is called a capacitive micromachined ultrasound
transducer or CMUT (Figure 4.9).
Cavities are micromachined into a silicon substrate that will resonate at ultra-
sound frequencies (Figure 4.10). A thin metalised membrane is suspended over
the cavity. The membrane and the silicon substrate then form the electrodes, and a
voltage bias is applied between them. If an alternating driving voltage is then applied
on top of the bias, the changing capacitance of the cavity will cause the membrane to
Element leads
Backing layer
Front electrode
PZT transducer
Quarter wave
matching layer
Front face
Beam from (elevation plane lens)
Sub-dicing single element
FIGURE 4.8 The basic structure of a piezoelectric transducer array.

CMUT Technology 67
Metallised membrane
Insulator
V
DC bias Vacuum cavity
Silicon substrate
FIGURE 4.9 Diagram of the main components of a CMUT transducer element.
100 μm
FIGURE 4.10 A scanning electron microscopy image of a CMUT silicon back plate showing a
grid of the resonant cavities. Source: Cianci et al. [2]/with permission of Springer Nature.
vibrate and transmit an ultrasound wave. When the membrane is caused to vibrate
by a returning echo, the capacitance of the cavity will change and an alternating
voltage will be generated forming the detected signal in reception. The resonant
frequency depends on the size of the cavity and the stiffness of the membrane. Many
of these capacitor cells may be connected in parallel to implement transducers and
arrays with different shapes and sizes and the required wide frequency response
(Figure 4.11).
The surface is covered with a thin layer of elastic polymer to provide a hygienic
insulating surface. A very large frequency arrange is achievable, e.g. 2–22 MHz within
one probe. The acoustic impedance of the vibrating membrane is close to that of
tissue, so no matching layer is required for efficient transmission into the body. A typ-
ical cavity size is 0.1–0.2 μm.
CMUT devices have a number of advantages. These are as follows:
• It is easier to construct 2-D arrays with CMUT technology using standard

microelectronic manufacturing technology.
• Large numbers of CMUTs in an array enable very wide bandwidths to
be achieved.
200 μm
FIGURE 4.11 A portion of a CMUT array of rectangular elements made of many CMUT
units (pitch: 250 μm × 12 mm). The electrode connections to each element are seen at the top.
Source: Cianci et al. [2]/with permission of Springer Nature.
• The wide bandwidth makes high frequency operation easier to achieve, e.g. for
harmonic or very high frequency – high resolution imaging.
• CMUTs have high sensitivity with low noise.
• No matching layer is required.
• Silicon technology integrates well with other electronic circuitry
1-D, 1.5-D, AND 2-D ARRAYS
The basic linear array is a one-dimensional array imaging in a single scan plane
(Figure 4.12).
A curvilinear array is a linear array bent round a circular arc to give a sector
shaped scan (Figure 4.13). The construction of a phased array is similar to that of
a linear array but with very close spacing of the elements so that they are <1λ apart.
The cylindrical lens shown on Figure 4.12 gives weak focusing in the elevation
plane to reduce slice thickness. This will give a general reduction in slice thickness,
particularly at the focal distance of the lens. Further improvement in slice thickness
has been achieved in two ways: the Hanafy lens and the 1.5-D array.
The Hanafy Lens

The Hanafy lens uses transducer elements that vary in thickness in the y-direction
to give improved slice thickness throughout the image rage (Figure 4.14). As it is a
Cylindrical
(a) lens (b)
FIGURE 4.12 Diagram of a linear array showing the elements and front face cylindrical lens.
FIGURE 4.13 Showing the construction of a Philips C5-2 128 element curved array transducer
inside the casing, behind the front lens facia. Source: Binarysequence [3]/Wikimedia Commons/
CC BY-SA 4.0.
PZT transducer
Matching layer
Resultant beam
x Low frequency focus
z High frequency focus
FIGURE 4.14 Diagram showing the focusing in the elevation plane of a Hanafy lens.
curved transducer, it will focus the ultrasound. Where the element is thin at its centre,
it will resonate with a higher frequency. As the distance along the y-axis away from
the centre increases, the resonant frequency decreases. The higher frequency from
the centre of the element will not penetrate so far into tissue so that region will focus
at shallow depths. The lower frequencies from a wider y-axis aperture will focus at
greater depth. Overall, there is an improvement in slice thickness throughout the
transducer range.
1.5-D Array
1.5-D array is also known as a ‘one-and-a-half-D’ array (Figure 4.15).
Each element is divided into three or more individual elements in the y-direction
(elevation). They can then perform some degree of electronic focusing in the elevation
plane to improve image quality.
2-D Array
2-D array is also known as a matrix array.
The matrix array has multiple elements in x- and y-direction (Figure 4.16).
Ultrasound beams may then be controlled both laterally and in elevation to give the
following advantages:
• The beam can be fully focused in both lateral and elevation planes giving
better resolution and greatly reducing slice thickness artefacts (see
Chapter 7).
• The beam can be steered in the x and y direction to give a 3-D data set from a
volume of tissue.
• Collecting a 3-D volume data set over a period of time enables 4-D scanning to
be performed, e.g. to show complete information of heart valve motion across
the cardiac cycle.
FIGURE 4.15 Arrangement of the elements of a so-called ‘one-and-a-half-D’ array.

FIGURE 4.16 A 2-D array.
A matrix array will typically have 2000–50 000 elements. 3D data acquisition pres-
ents significant hardware demands in addressing a large number of elements and
digitising and storing the information received. CMUT transducers lend themselves
well to the construction of 2D transducers as they can be fabricated using microelec-
tronic techniques. Sophisticated beamforming methods that use a reduced number
of elements to achieve the same results as using a full array are being developed,
sometimes called sparse arrays [4]. One example of such a technique is known as a
row-column array [5, 6]. In terms of firing, the elements are arranged as a set of rows
and columns as shown in Figure 4.17. By firing pulses from these two orthogonal
τy1
τy2
Σ Y
τy3
τy4
Ultrasound system
τx1 τx2 τx3 τx4
Processing Σ
FIGURE 4.17 A 2D row-column array showing the set of rows and set of columns firing as two
groups of elements. Source: Savoia et al. [5]/with permission of IEEE.
array sets in phase then in a second pulse firing the two arrays in opposite phase to
one another, then subtracting the two pulses, 3D images can be formed that effectively
use 2N elements in a probe instead of N2, where N is the number of elements in a row.
REFERENCES
1. Ergun, S., Zhuang, X., Huang, Y. et al. (2005). Capacitive micromachined ultrasonic transducer
technology for medical ultrasound imaging. Proceedings of SPIE - The International Society for
Optical Engineering 5750: 58–68.
2. Cianci, E., Foglietti, V., Minotti, A. et al. (2006). Fabrication techniques in micromachined capac-
itive ultrasonic transducers and their applications. MEMS/NEMS 2006: 353–382.
3. Source: Binarysequence (2020). Licenced under CC Attribution-Share Alike 4.0 International.
https://upload.wikimedia.org/wikipedia/commons/8/8d/Curved_Array_Ultrasound_Sensor_
Construction.jpg (accessed August 2022).
4. Provost, J., Papadacci, C., Arango, J.E. et al. (2014). 3D ultrafast ultrasound imaging in vivo. Phys-
ics in Medicine and Biology 59: L1–L13. https://doi.org/10.1088/0031-9155/59/19/L1.
5. Savoia, A., Bavaro, V., Caliano, G. et al. (2007). P2B-4 crisscross 2D cMUT array: beamform-
ing strategy and synthetic 3D imaging results. IEEE Ultrasonics Symposium Proceedings 2007:
1514–1517. https://doi.org/10.1109/ULTSYM.2007.381.
6. Jensen, J.A., Ommen, M.L., Øygard, S.H. et al. (2020). Three-dimensional super resolution imag-
ing using a row-column array. IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency
Control 67: 538–546. https://doi.org/10.1109/TUFFC.2019.2948563.
CHAPTER 5
Image Formation
The pulse-echo principle is the basis of nearly all diagnostic ultrasound techniques
(Figure 5.1). A pulse of ultrasound is transmitted into the body and echoes arrive back
at the transducer from ever greater depths with time after the pulse is transmitted.
To get information from the target, the pulse has to travel twice the target distance d,
i.e. out to, and back from the target, travelling at the speed of sound c in the medium,
e.g. tissue.
2d
Echo arrival time t
c
For soft tissue, c 1540 ms 1 then, t 13 s / cm go and return
e.g. if a target is 6 cm deep, then echoes arrive back at the transducer 13 × 6 = 78 μs

after transmit.
Ultrasound systems normally assume a speed of sound of 1540 ms−1 in order to
assign a depth to the origin of the echoes displayed in the image. This is known as
the system velocity. It is the average speed of sound in soft tissue, which in reality
varies by ±5% between different tissue types (Table 1.1). For some tissues, where the
speed of sound is uniform but different to 1540 ms−1, such as within the orbit of the
eye, some equipment will allow the user to alter the system velocity in order to make
accurate measurements.

73
74 Chapter 5 Image Formation
Echo
FIGURE 5.1 The pulse echo principle.
IMAGE MODES
The names used for the different types of greyscale image display come from the
development of radar where the same types of display were first developed.
A-Mode (Amplitude Mode)

An ‘oscilloscope’ type of display showing the amplitude of the echo signal from differ-
ent depths (Figure 5.2).
This type of display is not used much today but may be used in ophthalmology to
calculate the power of the eye where the cornea, lens, and vitreous humour all have
different speeds of sound. There are strong interfaces with echo free regions in bet-
ween (Figure 5.3).
If the A-mode information is displayed using a bright spot whose brightness
depends on the amplitude of the signal, then we produce the equivalent of one scan
line on an image display.
M-Mode (Motion Mode)

M-mode allows moving targets to be displayed so their movement can be appreciated
by the sonographer. Repeated pulses are fired along one single scan line and return-
ing echoes are displayed as a series of A-mode bright-spot scan lines laid next to each
other so as to show depth versus time. This gives an M-mode display where the change
in position of returning echoes from along the scan line can be seen (Figure 5.4).
Amplitude
Depth/time
FIGURE 5.2 An A-mode scan line displayed as an oscilloscope display of amplitude and a series
of spots whose grey level shows the signal amplitude.
Image Modes 75
FIGURE 5.3 A-mode of the eye with B-mode image for comparison. The image shows a dense
cataract and axial high-myopia. Note the separate speed of sound values used for anterior chamber
AC, lens L, and vitreous V indicated on the second line of results. Source: Silverman [1]/Originally
published by and used with permission from Dove Medical Press Ltd.
Time
Depth
(a) (b)
FIGURE 5.4 (a) Principle of an M-mode scan with (b) an M-mode across the left ventricle
showing tips of mitral valve motion. Source: Courtesy: Chris Eggett.
• Stationary reflectors appear at constant depth with time.

• Moving reflectors show their change in depth from the transducer with time.
The scan line used to form the M-mode display is indicated on the image display
and can be positioned to give information from the target of interest.
M-mode is primarily used in cardiac work to examine the movement of the heart
wall and cardiac valves.
Skin
Linear
image Sector
image
(a) (b)
FIGURE 5.5 The scan pattern of (a) a linear image and (b) a sector image.
B-Mode or Greyscale Image (Brightness Mode)

This is the standard imaging mode used for most ultrasound investigations.
There are two basic image formats: linear and sector formats. These are the
equivalent of moving the probe along the skin to gather an image slice or rocking the
probe about a fixed point on the skin to sweep through a sector (Figure 5.5).
Each format has areas of advantageous use, for example,
• Linear – For vascular work, to look at lengths of straight vessels, where the
angle of the probe surface to the vessel is maintained.
• Sector – Small footprint on the skin, but the image shape allows an extended
organ to be seen, e.g. imaging the heart with access between the ribs or to image
the extended length of a kidney.
The plane of the image through the body is known as the image plane or
scan plane.
LINEAR IMAGE FORMATION
Linear Array
A linear array is an ultrasound probe formed with a straight line of closely spaced
transducer elements (typical number, 160 elements). In order to achieve an aperture
of sufficient width to form a narrow beam, several adjacent elements are connected
electronically and fired together as one transducer (e.g. 8 elements). When the pulse-
echo sequence from that group is complete, the element at one end of the group is
dropped off and the next adjacent element at the other end is added to form a new
Step group along

to give successive
scan lines
Group fired to
give one scan line
FIGURE 5.6 The principle of stepping a group of active elements across a linear array.
group (Figure 5.6). The active aperture has then effectively stepped along the array by
one element from the previous group. This is then fired with a pulse-echo sequence
producing a beam that has moved across the front of the probe by a distance of one
element from the previous beam.
By continuing this process right across the array, a sequence of beams is produced
that sweeps across the width of the array. These beams obtain echo data across the
whole scan plane to form an image that is a slice through the tissue being scanned.
The whole scan plane can be scanned in the time it takes, say, 120 beams
to send out their pulse and obtain echoes from the depth of interest, e.g.
13 μs × 10 cm × 120 = 15.6 ms for a depth of 10 cm.
Frame Rate
Frame rate is the number of times the complete field of view is scanned per second.
In the example just given, this would produce a frame rate of 64 frames per second.
1
i.e. Frame rate 64 fps
15.6 ms
This is fast enough to appreciate movement in real time. In practice, the frame
rate would be somewhat slower than this to allow for echoes from the previous pulse
to die away and can be changed by the image acquisition techniques used as discussed
below and in Chapter 7.
Curvilinear Array
A curvilinear array uses the same method of stepping along a group of elements
to form an image as the linear array, but with the array formed into a curve to give
a sector shaped display, giving, for example, a wide field of view in the abdomen
(Figure 5.7).
Group fired to Step group along

give one scan line to give successive
scan lines
FIGURE 5.7 The principle of stepping a group of active elements across a curvilinear array.
FIGURE 5.8 The scan sector produced by beam steering in a phased array.
Phased Array
A phased array uses beam steering to form a sector shaped image using the whole
array as the active aperture for every pulse (Figure 5.8). By forming the probe with
very closely spaced elements, less than a wavelength apart, the beam can be steered
through a large angle without incurring the problem of grating lobes (see Chapter 3).
This enables the probe to have a small footprint on the skin but with the image plane
spreading out inside the body to image an extended organ. For example, imaging the
heart from between the ribs.
Extended Field of View

Extended field of view uses beam steering to increase the field of view of a linear
array at the extremities of the image (Figure 5.9).
Panoramic View
Another way in which the field of view may be extended is to store a sequence of
images as the probe is physically moved across the surface of the skin along the line of
FIGURE 5.9 Using beam steering to extend the field of view of a linear array.
the scan plane (Figure 5.10). The images are then electronically matched and blended
in a similar manner to the panoramic view setting on a camera. By this means an
image may be formed, say, of a large cyst or a blood vessel that is bigger or longer than
the field of view of a single image (Figure 5.11). The probe must be moved smoothly
across the skin by the operator, and the calibration of lateral distance on the image
will be poorer than for an image formed when the probe is stationary. The image is
built up over a period of time, as the probe is manually moved, so the image produced
is not a real-time live image.
Probe physically
moved along skin
Skin
FIGURE 5.10 The principle of producing an extended field of view by physically moving the
probe on the skin.
FIGURE 5.11 A panoramic scan of the aorta. Source: Courtesy of Simon Elliott.
3D IMAGING
Three-dimensional (3D) imaging of a block of tissue may be obtained in one of

three ways.
Hand Scanning
In the same manner that an extended field of view may be obtained in the scan plane,
by physically moving the probe across the skin in the scan plane, a 3D image may be
obtained by manually sweeping the probe across the skin in a direction perpendic-
ular to the scan plane (Figure 5.12). These images are then able to give a 3D data set
through a block of tissue. In this technique, the calibration of distance in the out-of-
scan plane axis (y-axis) cannot be relied upon. As with the panoramic view, the image
produced is not a real-time live image.
Mechanical 3D Probe
A curvilinear array is housed in a probe where it is made to oscillate by a motor in the
out-of-scan-plane direction as shown in Figure 5.12. A 3D data set is obtained. This
may be used for measurement as the direction the probe is pointing is known at all
times, so true distances may be calculated. Data acquisition is fast enough to give a
real-time display but with a relatively low frame rate [2].
2D Matrix Array
A two-dimensional (2D) electronic array of elements may be electronically scanned
in the two perpendicular planes, using beam steering, to produce a 3D data set in real
time (Figure 5.13).
FIGURE 5.12 Scanning a 3D volume of tissue by physically rocking the probe in the
elevation plane.
3D Imaging 81
FIGURE 5.13 Producing a volume scan by steering in the x-and y-plane with a 2D array.
3D Image Display
Having acquired a 3D data set, images may be displayed from any of the orthogonal or
other planes in the volume data set (Figure 5.14). For example, a set of slices through
an organ (Figure 5.15). This includes scanning in a plane that is orthogonal to the
direction of propagation of the ultrasound beam at constant depth from the trans-
ducer. Such a scan is in the coronal plane and has been called a C-scan.
FIGURE 5.14 Multiplanar view of an ultrasound volume of the foetal brain at mid-gestation
obtained with a transvaginal approach from the sagittal plane B. Source: Pilu et al. [3]/with
permission of John Wiley & Sons.
FIGURE 5.15 Multiple slices of the volume shown in Figure 5.15 orientated along the sagittal
plane. Source: Pilu et al. [3]/with permission of John Wiley & Sons.
CINE LOOP
Cine loop is the name given to the storage of a sequence of images that may be played
back for review. Where a 3D data set is recorded and stored to form a stored time
sequence, the term 4D scanning is used, with the fourth dimension being time.
ENDOPROBES
An endoprobe is one that can be introduced inside the body via a natural body cavity
or via a minimally invasive surgical wound, e.g. endovascular, oesophageal, rectal,
and transvaginal probe (Figure 5.16).
These probes may have a linear array, a tight curvilinear format or may be
formed from a set of elements around the probe to produce a radial image. In
a radial image, the ultrasound beam is swept round like the light from a light-
house (Figure 5.17). This produces an image plane like a radar view with the
Endoprobes 83
(a) (b)
(c) (d)
FIGURE 5.16 Examples of scan formats on endoprobes. Panels (a) and (b) have tight curvilinear
arrays as might be found in transvaginal probes, (c) has a linear array and a radial array as might
be found on a transrectal probe to examine the prostate, and (d) has two orthogonally arranged
phased arrays as used in a transoesophageal probe for echocardiography. Source: Whittingham [4]/
with permission of Elsevier.
Image plane
Probe
position
(a) (b)
FIGURE 5.17 (a) Illustration of a radial image produced by an endoprobe with radial elements.
(b) An example of a radial image showing plaque in a coronary artery. Source: (b) Nissen and
Yock [5]/with permission of Wolters Kluwer Health, Inc.
probe in the body cavity at the centre of the field of view. An example of this
form of imaging is intravascular ultrasound, which is used to examine diseased
arteries. By using a pull-back technique, a 3D data set of the surrounding tissue
may be obtained.
CHOOSING A PROBE
The choice of probe for a particular application will depend on a number of factors.
The imaging target will be the prime factor and will usually dictate whether a linear,
sector, or endoprobe format is used. There may be constraints on the footprint of the
probe for access into awkward angles of approach to a target, e.g. the use of a phased
array for cardiac work accessing between the ribs, or a tightly curved curvilinear
array for access to a neonate fontanelle. For peripheral vascular work or small parts
scanning a linear array may be most appropriate.
The second factor is image resolution. Here, consideration must be given to
resolution versus penetration to the target of interest. The probe with the highest fre-
quency that allows the target of interest to be adequately imaged should be chosen.
Where 3D data are required a probe with that ability must be used.
Resolution will also be affected by the focusing used and the frequency used
within the probe bandwidth.
FOCUSING
We saw in Chapter 3 that focusing the ultrasound beam gives an improvement in

beamwidth and hence resolution at the focal point. Outside the focal zone, the reso-
lution is poorer as the beam spreads out. Focusing to several depths is called multi-
zone focusing. This can be achieved using electronic focusing and is used to improve
the resolution at all depths in the image.
Successive pulses do not have to have the same shape as each other. The focal
depth between each pulse can be changed. Neither do the transmit beam and receive
beam have to have the same shape. Using this fact, resolution can be improved across
the whole image by choosing appropriate focal zones for each pulse.
Transmit Focus
This is the focus used for the transmitted pulse.
Once a pulse has left the transducer, no more processing can be done to it. Therefore,
each transmitted pulse can only be focused to a single depth. In other words, the focal
depth is static for each pulse echo sequence. To focus at a different depth along that scan
line requires another transmitted pulse focused to the new depth. In order to focus to three
different depths in transmit, it will require three transmit pulses (Figure 5.18). For each
transmitted focal zone, echoes are only detected for the depth at which the beam is focused.
Doing this for each scan line in the image will reduce the frame rate by a factor of 3.
Receive Focus
This is the focus used for the received echoes.
On receiving, we can rapidly switch beam shapes for echoes coming from dif-
ferent depths. For example, we can focus to 2 cm deep when echoes are coming
Focusing 85
f1
f2
f3
a3 a2 a1
f1 f2 f3
FIGURE 5.18 Using different apertures and delays to produce a set of focal zones at
different depths.
from 2 cm deep and then switch to a receive focus of 4 cm deep when echoes are
arriving from 4 cm deep, etc. In other words, the focal zone can be set for each suc-
cessive depth for which echoes are returning. By switching in, say, 10 focal zones in
quick succession we can construct a very narrow receive beam (Figure 5.19). This
is known as dynamic receive focusing as the focal zone changes within a single
pulse-echo sequence.
All machines will automatically use dynamic multi-zone focusing for the receive
beam so as to give the best resolution possible.
For the transmit focus the user must select the depth of the focal point. This
should be at the target of interest so that the target is imaged with the finest resolu-
tion. The user can choose to use more than one focal zone on transmit with the pen-
alty of reduced frame rate.
NOTES
The overall resolution of the image will result from the combination of the transmit
and receive beam shapes multiplied together to produce the image.
There are techniques whereby dynamic transmit focusing can be realised,

extending the focal zone within the image. The most effective way of achieving this is
through synthetic aperture techniques described in Chapter 13.
Receive beam shape
1 2 3 4 5
FIGURE 5.19 The narrow beam produced by using a number of focal zones in receive.
INCREASING FRAME RATE
Conventional B-mode imaging frame rates are limited by the need to wait for
the go-and-return time for pulses for each beam in the image. Frame rate may
be increased by using multiple beams to increase the rate of acquisition of
image data.
One way to achieve this is to transmit a wider beam than normal and receive the
echoes through 2 or 4 beam formers to create 2 or 4 receive beams for each transmit
beam. This will be at some cost to lateral resolution but will increase frame rate by a
factor of 2 or 4. Alternatively, multiple beams can be transmitted and detected simul-
taneously at positions where they do not interfere with one another. Using such tech-
niques, real data frame rates of 60 fps can be achieved.
Synthetic Aperture Imaging

Synthetic aperture imaging is a sophisticated technique whereby an ideal ultrasound
source is mimicked by firing elements in an array so as to produce a beam that would
have come from the ideal source. By creating such sources and combining the received
echoes appropriately, transmit and receive focuses can be synthesised for every point
in the field of view. This can be achieved at very high speed with greatly improved
frame rates. It also improves contrast and resolution throughout the image as every
point is focused on transmit and receive.
Chapter 13 describes in detail how synthetic aperture imaging works.
NOTE
Some scanners using these techniques do not require the user to set the transmit
focal depth.
USER CONTROL
Resolution/Penetration
Wideband transducers allow a range of frequencies to be transmitted and detected.
The user has some control over the frequencies used to form the image. This is
often presented in the form of a control for ‘Resolution/Penetration’. This may be
adjusted by the user to obtain the best image. It is trading frequency for penetration
within a single probe.
Another technique to improve resolution by using higher frequencies is
harmonic imaging.
User Control 87
Background Theory to Understanding Harmonic Imaging
1. Superposition of Waves
If two waves coincide then the amplitude at each point is the sum of those
two wave’s amplitudes at that point, e.g. a peak plus a peak gives a bigger peak,
a peak plus a trough cancel each other out as seen in Chapter 2.
Figure 5.20 shows the superposition of two waves where one wave has
exactly twice the frequency of the other.
NOTE
If the phase of 2f0 relative to f0 changes then the shape of the resultant wave will
also change.
2. Harmonics
If the frequency of a given sine wave f0 is increased by an integer multiple,
i.e. ×2, ×3, ×4, etc., we generate a set of harmonic frequencies. In music, a
doubling of frequency gives a note one octave higher (Figure 5.21).
f0 is called the fundamental or first harmonic.
2f0 is called the second harmonic.
3f0 is called the third harmonic.
4f0 is called the fourth harmonic.
Harmonics can be heard on a stringed instrument, or seen by shaking a skipping
rope, by forcing the vibrations to divide the string into half, third, etc. (Figure 5.22)
The French mathematician Fourier showed that any repeating waveform
can be shown to be made up of the superposition of a fundamental + harmonics
of higher multiples of f0.
f0
2f0
sum
FIGURE 5.20 The superposition of a wave of frequency f0 and a frequency of 2f0.
(continued )
(continued)
f0
2f0
3f0
FIGURE 5.21 The fundamental and first two harmonics of a wave of frequency f0.
Fundamental
Second harmonic
Third harmonic
FIGURE 5.22 The first three harmonics as seen on a vibrating string.
Fundamental
Fundamental
+ 3rd harmonic
FIGURE 5.23 Diagram showing how a square wave is built up of odd harmonics.
The higher harmonics usually have smaller amplitudes.

Example: A square wave (repeating) is made up of the fundamental + odd
harmonics (3,5,7,9. . ..) (Figure 5.23).
The more harmonics that are added, the closer we get to a square wave.
Adding all the odd harmonics gives a pure square wave
Here, we can see that it is the higher harmonics that turn a sine wave into
a square wave. It is generally true that if a waveform has sharply rising/falling
edges, then there must be higher harmonics in its makeup.
SUMMARY
• Any repeating signal can be broken down into a set of harmonic frequencies
superimposed on one another – the amplitude and relative phase of each
harmonic component must be right (like a cooking recipe – too much salt
does not work).
• f0 is the fundamental frequency and first harmonic. Harmonics are multi-
ples of the fundamental.
• If a waveform has sharp edges, then it must include higher ‘faster moving’
frequencies in its composition.
ULTRASOUND HARMONICS
Up to now, we have assumed that a pulse travels through tissue without any change
to its shape, and it behaves in what is called a ‘linear’ way. That is, if you increase the
intensity the wave has a bigger amplitude but still has the same shape. In reality, as a
wave or pulse travels through tissue it becomes distorted due to non-linear propaga-
tion. This occurs because the speed of sound at peak pressure is slightly higher than
the speed of sound at minimum pressure. The result is that the peaks move faster than
the troughs and tend to catch them up (Figure 5.24). This is like the peaks of sea waves
getting steeper and breaking on a beach as the sea gets shallower. The peak catches
up with the next trough.
Question: What happens to the spectrum (frequency content) of the wave as it

changes shape?
Answer: Second (+other) harmonics are generated. As we saw in the theory
above, sharper edges imply that higher frequencies must be present.
Energy goes from the fundamental into the second harmonic, i.e. non-linear prop-
agation (Figure 5.25). The Spectra shown in Figure 1.18 show the second harmonic in
a B-mode and a Doppler pulse due to non-linear distortion in the pulses.
We know that higher frequencies give better resolution, so if we can use the sec-
ond harmonic, so generated, to form our image, the resolution of the image will be
Peaks move faster

than troughs
FIGURE 5.24 The effect of non-linear propagation causes the peaks to get sharper and the
troughs to become wider.
f0 f0 2f0
FIGURE 5.25 The energy spectrum of a wave showing non-linear behaviour. Energy is lost from
the fundamental f0 and goes into the generated second harmonic 2f0.
Harmonics
generated
FIGURE 5.26 Showing the area within the beam where intensities are great enough to generate
harmonics. Source: Adapted from Whittingham [4].
improved. Using the second harmonic to form the image is called harmonic imaging.
In addition to higher frequencies giving improved resolution, the ‘beam’ produced at
the harmonic will be narrower. This is because harmonics are produced at higher
amplitudes and the highest amplitudes are seen along the main axis of the beam, as
shown in Figure 5.26. This ‘harmonic beam’ has no side lobes, so clutter is also reduced.
Aside: Filters
High-Pass Filter – A device that allows high frequencies above its cut-off fre-
quency to pass, and blocks low frequencies.
Low-Pass Filter – A device that allows low frequencies below its cut-off fre-
quency to pass and which blocks high frequencies.
These devices may be electronic devices or be implemented as a software algo-
rithm to process a digitized signal.
Harmonic Imaging
There are two basic ways we can obtain the harmonic information to form an image.
Using a filter method (Figure 5.27):
A wideband transducer will enable both the fundamental frequency to be trans-
mitted and the second harmonic to be detected. If a high-pass filter is used to cut out
f0 from the received signal, then the image can be formed from 2f0. Slightly longer
pulses than for normal B-mode are used to narrow the pulse bandwidth and improve
High pass
filter
T × f0 R × 2 f0 f
FIGURE 5.27 The use of a high pass filter to detect the second harmonic signal within the
transducer bandwidth.
B A + B gives C
C
No harmonics Harmonics generated
FIGURE 5.28 The pulse inversion method of detecting the harmonic signal. The echoes from
two consecutive pulses A and B are added together to give the resultant signal C.
separation of first and second harmonic for filtering. This will slightly degrade the
axial resolution. But overall there will be an improvement in resolution from using
the higher frequency harmonic to form the image.
Using a pulse inversion method (Figure 5.28):
This is the more common implementation and uses normal length short pulses.
It requires two transmit pulses of opposite phase to be transmitted in sequence. The
echoes from the two pulse sequences are stored and added together. At the fundamental
frequency where the pulse maintains its shape, the two echo trains cancel each other
out. Where harmonics are generated, there will be a net signal that can be used to gen-
erate a harmonic image. Axial resolution is maintained at a cost of frame rate.
Pros and Cons of Harmonic Imaging

Pros
• Better resolution with a higher frequency and narrower beams.

• Removes ‘clutter’ from the image (due to reduction in sidelobes, reverberations
and multiple echo paths)
(a) (b)
FIGURE 5.29 Image of renal angiomyolipomas showing the decluttering effect of harmonic
imaging. Image (a) has no harmonic processing and image (b) has harmonics activated. Source:
Khandelwal et al. [6]/OAE Publishing Inc., CC BY 4.0.
Cons
• Weaker echoes at 2f0 with higher attenuation so penetration is poorer.
In practice, the harmonic imaging mode on a scanner uses a combination of

second harmonic and first harmonic to give an overall improvement to the image
without losing penetration. Its effect is seen in Figure 5.29.
Harmonic Imaging is something that the user can choose to turn on or off.
CODED EXCITATION
Coded excitation, sometimes called pulse compression, is a technique used

to improve sensitivity to weak echoes. Echoes from the very short pulses used for
imaging are hard to detect in the presence of significant background noise and clutter.
Very short pulses ‘look like’ noise spikes. One way to improve signal-to-noise ratio
(SNR) would be to increase the transmitted power but that has safety implications. By
imposing a ‘code’ on to the transmitted pulse, the echoes can be decoded and the true
echo signal recovered from a noisy background.
Instead of a short high intensity pulse, a longer pulse with lower average intensity
that carries a code is used. The longer pulse would give poorer resolution except that
when it is decoded with a complimentary code, the result is a compressed output of
the same length as a normal short pulse. This can give decoded pulses with a larger
peak intensity than would have been obtainable from a short pulse within safe inten-
sity limits. Hence, the term ‘pulse compression’ is used.
Coded Excitation 93
Examples of Coded Excitation
There are a number of ways a pulse can be encoded but to give an idea of what
is involved the following examples are given:
Transmit two pulses which are shaped with a pseudo-random code A and a
complementary code B. Their echoes are received. To decode, the echoes from
each pulse are matched to a time reversed copy of the transmitted pulse in an
cross-correlator as shown in Figure 5.30 (see Chapter 9). The correlated outputs
are then summed together. The result is a strong signal for the encoded echoes
whilst noise in the signal tends to cancel out.
The symbol ⊗ means cross-correlation.
The encoding may use changes in the phase of the signal to give a binary
+1, −1 code, or the encoding may be produced by changing the transmit driving
frequency during the course of the pulse.
In the example in Figure 5.31, the code is shown as changes in phase
within the transmitted pulse and the code used only requires one transmission-
decoding sequence.
Coded pulse 1 1 1 1 1 = 1 1
4
Σ
2
1 1
=
Coded pulse 2
–1 –1 –1 –1 Sum1 and 2
Transmit Decoder Output
FIGURE 5.30 Diagram demonstrating the principle signal detection using coded excitation.
+ + + + + + + +
– –
A B
FIGURE 5.31 Coding of the signal using 180° phase shifts and the detected signal.
Chirp Pulse
A chirp pulse is a pulse whose frequency varies throughout the pulse such that
the driving frequency is swept through a range of frequencies from the start to the
end of the pulse (Figure 5.32). A matched filter in the receiver ensures that only
echoes matching the chirp are detected enabling true echoes to be discriminated
from background noise.
FIGURE 5.32 Example of a chirp pulse covering a range of frequencies as the frequency
increases throughout the pulse.
REFERENCES
1. Silverman, R.H. (2016). Focused ultrasound in ophthalmology. Clinical Ophthalmology 10:
1865–1875. https://doi.org/10.2147/OPTH.S99535.
2. Huang, Q. and Zeng, Z. (2017). A review on real-time 3D ultrasound imaging technology. BioMed
Research International 2017: Mar 26. https://doi.org/10.1155/2017/6027029.
3. Pilu, G., Ghi, T., Carletti, A. et al. (2007). Three-dimensional ultrasound examination of the fetal
central nervous system. Ultrasound in Obstetrics & Gynecology 30: 233–245.
4. Whittingham, T.A. (2007). Medical diagnostic applications and sources. Progress in Biophysics
and Molecular Biology 93: 84–110.
5. Nissen, S.E. and Yock, P. (2001). Novel pathophysiological insights and current clinical applica-
tions. Circulation 103: 604–616. https://doi.org/10.1161/01.CIR.103.4.604.
6. Khandelwal, N., Chowdhury, V., and Gupta, A.K. (2013). Diagnostic Radiology: Recent Advances
and Applied Physics in Imaging. Jaypee Brothers Medical Publishers. ISBN 10: 9350904977.
CHAPTER 6
The B-Mode Scanner
Looking at the basic layout of an ultrasound B-mode imaging system gives an overview
of the basic processes involved in producing an ultrasound image and shows
where the main controls accessible to the user are enabling them to interact with
the image forming process. We begin with the transmission side and then look at the
receive side of a scanner.
TRANSMISSION SIDE OF A SCANNER
The transmission side of a B-mode scanner is shown in Figure 6.1.

Modern scanners can:
• Alter the shape of the transmit (Tx) pulse.

• Change the transmit power between individual elements.
• Change the timing of the pulses between elements and control the phase of the
pulses at each element.
• Change the frequency of the pulse.
• Change the aperture used.
and it can do all of these using different values from one pulse to the next, i.e. no two
pulses need be the same.

95
96 Chapter 6 The B-Mode Scanner
Focal depth Tx power
td
Pulse generator
Beam
(shape) Tx
former
(timing)
FIGURE 6.1 Block diagram of the transmit side of B-mode scanner.
This enables:
• Dynamic multi-zone focusing

• Beam steering
• Beam apodization to be used to suppress side lobes and reduce ‘clutter’ in the image.
• Control the phase of the pulse for harmonic imaging
• Simultaneous B-mode and Doppler imaging using different transmit frequencies
• Allow complex data acquisition, combining echoes from multiple apertures to
achieve uniform focusing throughout the image and faster frame rates.
• Generate encoded wave trains – coded excitation – to improve signal to noise
ratio of weak signals.
Beam Former
The beam former controls exactly which pulse – shape, timing, phase – is produced
by each element in the transducer to give the required focal depth and beam steering
for each transmitted ultrasound pulse.
PRF – Pulse Repetition Frequency

This is typically 1000s of pulses per second. (recall: the go and return time to 10 cm
depth is 130 μs which is equivalent to a PRF of 7690 pulses per second)
USER CONTROLS
Transmit Power
Transmit power needs to be adjusted to minimise risk of injury from energy deposited
in the tissue whilst giving adequate penetration to image the target of interest.
Depth of Focus
The depth of focus needs to be set at or just below the target of interest. More than one
focal zone may be chosen to extend the focused field of view at the expense of frame rate.
Receive Side of a Scanner (Rx) 97
RECEIVE SIDE OF A SCANNER (Rx)
The receive side of a B-mode scanner is shown in Figure 6.2.
Beam Former
In receive the beam former sums the received echoes from each of the transducer
elements with appropriate time delays so as to form the ‘receive beam’ with the correct
focuses and steering.
Gain
As can be seen from Figure 6.3, the size of echoes has a very large range. In general,
the echo signal from within tissue, carrying the information we want to see, is very
weak and the signal must be amplified. Gain is another word for amplification.
Beam
Rx TGC RF detection
former
Image Image Scan

Display ADC
storage processing converter
FIGURE 6.2 Block diagram of receive side of B-mode scanner.
Size of echoes
1V 0 dB Soft tissue - air
100 mV –20 dB Liver - fat Large interfaces
10 mV –40 dB Liver - muscle
1 mV –60 dB Liver
100 μV –80 dB Scattering within tissue
10 μV –100 dB
Blood
Factor of 105
FIGURE 6.3 Size of echoes received from different tissue interfaces (note: the voltage is signal
amplitude, decibel values are for signal power).
User Control – Gain

• The overall signal gain may be controlled by the user. This will lift or depress
the overall brightness of the image on the display.
• If the gain is set too high, mid to high level echoes will be pushed towards peak
white on the display and information may be lost. The image will also become
noisy as low-level noise is over amplified.
• If the gain is set too low information in lower level echoes may be lost due to
under amplification.
• The gain should be adjusted so that the peak signal in the image just shows as
peak white whilst the lower level echoes that give the clinical information are
all distinguishable on the grey scale.
Time Gain Compensation

We saw in Chapter 2 that attenuation causes returning echoes strength to decrease
with distance from the transducer. The received signal therefore needs to be amplified
more and more to compensate for the loss in signal strength. This process is known
as Time Gain Compensation (TGC) – also known as ‘swept gain’ or Depth Gain
Control (DGC).
Without correction, the image would be bright near the surface and get gradu-
ally dimmer for greater depths in tissue as returning echoes get weaker due to atten-
uation from absorption, scattering, and beam divergence. To be clinically useful, the
image from a large organ, such as the liver, should have a uniform brightness at all
depths so that no clinically significant changes in appearance are missed. For each
transmit pulse, Gain that increases with Time of echo arrival is applied to Compen-
sate for attenuation of the pulse, i.e. Time Gain Compensation (or TGC). If the
correct amount of gain is applied, the signal strength from a uniform medium will
remain constant for all depths over the usable depth of penetration of the probe as
shown in Figure 6.4.
Received
RF signed
TGC gain t = depth
TGC output
FIGURE 6.4 The effect of Time Gain Compensation on received signal amplitude.
SUMMARY
• Due to attenuation, the received echo signal will reduce in amplitude for
echoes arising from further away from the transducer. On average, the level
falls off exponentially with depth of target.
• The TGC gain curve is applied to the returning echoes from each
transmit pulse.
• Applying TGC gives a uniform signal over the useable range of echoes. At
large depth no detectable echoes return and only electrical noise in the
system is amplified. The depth at which this happens defines the depth of
penetration for a particular transducer.
• Most machines will apply a basic TGC to returning signals by default.
• Fine adjustment must be controlled by the user via a set of sliders or other
controls that adjust the gain at various depths in the image so as to give a uni-
formly bright image over the useable penetration of the ultrasound probe.
• Some machines use sophisticated algorithms to improve image uniformity
automatically based on soft tissue speckle statistics.
(a) (b)
FIGURE 6.5 Images showing (a) poorly adjusted TGC losing information in the proximal and
distal zones and (b) the same scan correctly adjusted.
An example of how to adjust TGC correctly is given in Figure 6.5.
Analogue to Digital Converter – ADC

The ADC converts the analogue signal from the echoes into a digital signal
where each point in the signal is represented by number that can be stored in a
digital memory.
How Analogue to Digital Conversion Works
Example of a time varying signal (Figure 6.6):

Following digitisation the signal is described by the number sequence 3, 5,
7, 8, 7, 6, 7, 8, 9, 10, 8, 5, 1.
In each time interval T1, T2, T3, … the analogue value of the amplitude of
the echo signal is converted to the nearest integer value and represented by that
number. The complete echo train is then stored as a sequence of numbers (in
binary format).
NOTE
There is some loss of information as only integer values are stored, e.g. there is no
value for 5.7, it will be converted to 5.
The quality of the data therefore depends on A–D rate (how many times per
second a value is calculated) and the number of binary bits the signal is digitised
to. The amplitude of an echo signal from a 10 MHz transducer will change at
least 20 M times per second (going positive then negative).
Digitisation must therefore be carried out at least 20 M times per second in
order to give a faithful representation of the signal in digital form. Many scan-
ners will digitise the raw echo signal at 40 MHz.
Using binary representation of the digitized number each binary digit can
be 0 or 1. The number of bits used to represent each digitization point of the
signal will determine how the greyscale image will appear to the viewer.
Number of bits to convert to:
1 bit 0 and 1 = black or white only

2 bits 00,01,10,11 = black, white and 2 grey levels in between
8 bits 28 = 256 grey levels
12 bits 212 = 4096 grey levels – the human eye cannot distin-
guish this number of levels but
it results in a much smoother
looking image.
10
8
6
Signal
amplitude 4
2
t
3 5 7 8 7 6 7 8 9 10 9 5 1
FIGURE 6.6 Digitisation of an analogue signal.

Advantages of Digitising
• No further risk of introducing noise or distortion into the signal

• Powerful computing can be done = image processing
• Easy storage in memory
Echo Detection
This is the process whereby the raw echo signals are detected and turned into image
data. The raw signal is called the RF signal (for radio frequency). It is a time varying
signal whose phase and amplitude are constantly changing. The B-mode image
displays the amplitude of the echo arising from each point in the field of view.
NOTES
• The amplitude of the analogue signal may be directly detected and used to
form the B-mode image.
• Modern scanners often digitize the raw RF signal of the returning echoes,
preserving phase information to allow advanced signal processing tech-
niques to be used. The digital amplitude of the processed signal is then used
to form the B-mode image.
Detection of the Analogue RF Signal
The echo train from a transmitted pulse arriving at the transducer is a complex
signal with varying amplitude from many reflecting targets in the body (Figure 6.7).
The signal itself typically varies at the transmit frequency of the transducer. The
size of echoes arising from targets in the tissue will change the amplitude of the
echo signal as seen in the outline or envelope of the echo signal (Figure 6.8).
Consider One Pulse and One Reflecting Target
The echo signal is rectified (made all positive) and passed through a low
pass filter. This cuts out all the high frequency MHz components leaving the enve-
lope of the shape of the echo pulse. Rectification and low pass filtering together
is called envelope detection and together they produce a demodulated echo.
t
V
FIGURE 6.7 Example of the RF echo signal received at a transducer.

(continued)
(continued)
+ Pulse
t
V
– Demodulated
Rectification
echo
Envelope + Low pass filter
FIGURE 6.8 The principle of envelope detection with rectification to form a video pulse.
V Thresholds
FIGURE 6.9 Thresholds applied to limit signal levels.
The signal may be scaled and cut-off thresholds may be applied to remove low
level noise and limit the signal to a useable range as shown in Figure 6.9.
DYNAMIC RANGE AND TRANSFER FUNCTION

(GREYSCALE MAPPING)
Dynamic Range is the difference in signal strength between the strongest signal and
the weakest signal detected or displayed.
Contrast is the difference in grey level between adjacent targets in the image. For
example, the appearance of an image as black and white between adjacent areas is a high
contrast image versus subtle changes in grey level which would be a low contrast image.
At the low end of the range of signal strength will be ‘noise’. Some of this noise
will be from extremely low-level echoes but it will also include electrical noise from
within the scanner. The scanner may include a noise rejection control to remove this
very low-level signal that does not contribute to the useful information in the signal.
This is essentially a filter that blocks all signals below a chosen amplitude level.
The dynamic range can be controlled, limiting the size of signals to a narrow or
wider output range. This process of limiting the signal to a given range is called signal
compression.
NOTE
The receive gain used also interacts with the dynamic range by lifting or lowering
the echo levels within the dynamic range. For example, if the gain is set very high the
noise at the lower end of the echo signal range will extend over a greater proportion
of the dynamic range and less bright echoes will get pushed to toward the top white
level of the range. The overall appearance of the image will be brighter but the image
will be more noisy and important information may be lost.
TABLE 6.1 Showing How the Number of Bits Converts to Dynamic Range
Bits Value Range Dynamic Range (dB)
4 0–15 24
8 0–255 48
10 0–1023 60
12 0–4095 72
Using the decibel scale (see Appendix B) for a digitized signal (binary 0,1 bits),
we can see how the dynamic range relates to the number of bits a signal is digitized
to (Table 6.1).
As shown in Appendix B, doubling of amplitude = +6 dB

so 4 bits = 2 × 2 × 2 × 2 = 16 levels or 4 × 6 dB = 24 dB etc.
Dynamic Range
As shown in Figure 6.10, a wide dynamic range with a large number of grey levels will
give a ‘smooth’ appearance to the image. A narrow dynamic range with more signal com-
pression will give a more ‘contrasty’ image with a more black-and-white appearance.
• Echoes from the body have a dynamic range of 100 dB

• The human eye has a dynamic range at one time of ~30 dB (i.e. without adjust-
ment over time to dark/bright environment etc.)
• A good quality flat screen display monitor has a dynamic range of 60 dB
Transfer Function or Grey Scale Mapping

We therefore need to use signal compression to ‘fit’ the large dynamic range of the
echoes into the dynamic range of the display monitor. The relationship between the
input and output signal is called the Transfer Function or Grey Scale Mapping.
(a) (b) (c)
FIGURE 6.10 Showing the effect of increasing dynamic range (a) 36 dB, (b) 69 dB, (c) 90 dB.
An example of such a transfer function would be to scale the dynamic range of

the echo signal to fit the range of the display monitor using a linear scale as shown in
Figure 6.11.
Such a linear grey mapping gives equal emphasis to every change in grey level in
the image. Although this may seem a good idea, it is not usually the best choice for
diagnostic scanning. Clinically the most useful echoes are those in the mid-range of
amplitude. Very strong reflectors such as bone or gas filled structures appear as bright
white in the image and are clearly seen. At the bottom end, the signal becomes noisy
and it is using valuable grey scale span to emphasise these poor echoes. Therefore, a
mapping that emphasises the mid-range is a good way of distributing the diagnos-
tically most useful echoes across the available range of grey levels. With reference
to Figure 6.12, this leads to an ‘S’ shaped transfer function (a). Such a mapping de-
emphasises the very low-level echoes and the brightest echoes and emphasises the
mid-range echoes. Another choice would be to emphasise the mid-low level echoes
(b) where subtle changes in tissue texture may be seen, such as when looking for liver
metastases.
72 dB
Digital
0
–100 0 dB
Echo
FIGURE 6.11 Signal compression using a linear transfer function.
1023 1023 1023

Digital
Digital
Digital
0 0 0
–100 0 dB –100 0 dB –100 0 dB
Echo Echo Echo
(a) (b) (c)
FIGURE 6.12 Examples of transfer functions. (a) An “S” shaped curve emphasising the mid-
range echoes, (b) a curve emphasising low level echoes and (c) a step curve giving a binary black
and white image.
Linear S curve
(a) (b)
FIGURE 6.13 Illustrating a change in transfer function from a linear curve (a) to a S-shaped
curve (b).
The choice of mapping can dramatically alter the appearance of the image. Even
if a wide dynamic range is used, the choice of a steep mapping can produce a very
‘contrasty’ image. For example, in the extreme case, the step mapping shown in (c)
would give a black and white binary image. Every signal to the left of the vertical line
would show as black and every signal to the right would show as white. The level of
that black-white cut-off depends on where within the dynamic range the cut-off is
placed. In practice such a binary cut-off would never be used, but to think about it
illustrates the effect the mapping has relative to the dynamic range used. Examples of
linear and S-shaped transfer functions are shown in Figure 6.13.
SUMMARY
• Controlling dynamic range determines the compression of the echo signal
and how many grey levels may be used to display the image.
• A smaller dynamic range will make the image appear more ‘contrasty’ and
less smooth. Fine degrees of detail will be lost.
• The greyscale map used determines how the range of echoes, within the
dynamic range, are distributed across the greyscale displayed.
• By using a non-linear mapping, the majority of the greyscale may be used
to display the clinically useful information. Low-level noise and high-level
bright echoes are de-emphasised.
• When using an ‘S’-shaped mapping, the steepness of the curve will make
the image more or less contrasty, with the contrast cut-off level determined
by the positioning of the steep gradient within the dynamic range.
CONTRAST RESOLUTION
Contrast Resolution is the ability to detect a target whose grey level is different from
the surrounding image grey level. It is discussed in Chapter 7.
USER CONTROLS
Many scanners enable the user to control the dynamic range and which mapping
is used. The level to set the dynamic range and the mapping to choose depends on
a number of factors. Some scans benefit from a more ‘contrasty’ image, e.g. some
vascular scans or scans where it is important to see boundaries and interfaces within
tissue. Other images benefit from a ‘smooth’ image with a wide dynamic range, e.g.
looking for subtle changes in a diseased liver or when looking for fine detail.
As already noted, the main signal gain interacts with the dynamic range and
must be adjusted to optimise the visualisation of clinically useful information in
the image.
Some sonographers have a natural preference for a more or less contrasty image.
It is the case that your ‘eye-brain’ does get used to the image it regularly sees and will
pick up the necessary information from an image that another sonographer would find
less helpful. This partly explains why one person prefers the image of one scanner over
another – you get used to what you are familiar with. However, it is very important to
use a dynamic range and mapping that does not hide or minimise the information you
need to see for a particular scan.
One of the things a machine preset (see below) for a particular scan type does, is
to set a dynamic range and greyscale map for that scan. It is important for you as the
sonographer to ensure that the dynamic range and mapping is suitable for the patient
in front of you and to adjust them if necessary. A machine preset is not necessarily the
optimal setting.
IMAGE MEMORY
The image memory acts as a scan converter that converts an image from one format
to another. For example, write to the memory in a sector format, addressing pixels to
match acquisition of scan lines and then read out the memory in a rectilinear video
display format at video rate (Figure 6.14).
FRAME FREEZE
The image may be frozen to examine it, make measurements and move to long
term storage.
Read and Write Zoom 107
Video pixel
Scan line
FIGURE 6.14 Map of the image memory showing the relationship between the acquired image
lines and the rectilinear display format sent to the viewing monitor.
NOTE
Beware of introducing blurring in an image by moving the probe as you press ‘freeze’.
READ AND WRITE ZOOM
It is useful to be able to zoom in on a part of the image. Depending on how it is done,

it may or may not give more information. The zooming may be done as read zoom or
write zoom (Figure 6.15).
Read Zoom
The image is frozen and stored in memory and part of that stored image is read out
and enlarged to display on full screen. No new echo information is produced but fine
details may be clearer, e.g. for making measurements.
Write zoom
from new
image data
Read zoom
from image
memory
FIGURE 6.15 Illustrating the difference between write zoom which uses new scanning data and
read zoom which expands the image stored in memory.
Write Zoom
Write zoom uses new echo information to form the zoomed image. Instead of scanning
a full field, the ultrasound scanner just sends pulses into a restricted field of view. This
restricted field of view is then used to fill the whole display. Possible benefits are:
• Increased frame rate as there are fewer lines to interrogate.
or
• More closely spaced scan lines into the smaller field of view. This preserves
frame rate but reduces the need to interpolate so much (see Chapter 7). When
displayed, this ‘zoomed’ area has more real information as more pulses were
used to cover the field of view.
Remember – read zoom displays enlarged frozen data, write zoom displays live data
at a larger scale.
IMAGE PROCESSING
A number of image processing techniques may be applied to the image data.

These are aimed at improving the quality or appearance of the image and to
increase the diagnostic information available by bringing out certain features in
the image.
They may be applied as pre-storage or post-storage processing. That is, whether
the processing is performed on the data before it is stored in memory or whether it is
performed on the stored data.
Image processing is fully discussed in Chapter 7 and section “Advanced Image
Processing” in Chapter 12.
USER CONTROL
Exam Presets
The manufacturers of scanners provide the user with presets for the examination of
particular clinical areas of the body, e.g. small parts, abdominal, foetal exam etc.
NOTE
When activated each preset will change a wide range of controls including power
output, grayscale mapping, frequency, image processing etc.
User Control 109
In addition to presets provided by the manufacturer there is usually the option to

set up your own presets and store them for future use. It is important that as a user,
you know how to adjust the scanner controls to give a clinically useful image
yourself, without relying on the presets. Presets, whether the manufacturers or
your own, provide a quick way to progress an examination but you should take the
time to find out what each preset is adjusting and then make your own adjustments
in the light of that information. It will also be necessary to make adjustments during
an examination so that your machine is giving the optimal image for the patient in
front of you.
Measurements
Using the image stored in the memory, cursors may be superimposed to make mea-
surements. Measurements are discussed in detail in Chapter 10.
Figure 6.16 shows a block diagram of a full B-mode scanner.
NOTE
The order of some processes may differ on different machines.
Beam
Tx
former
Pulse
generator
Beam
Rx TGC ADC
detection
Pre-storage Image
Demodulation Display
processing memory
FIGURE 6.16 Block diagram of a complete B-mode imaging system.

CHAPTER 7
Image Quality and Artefacts
As with all medical imaging, image quality is an important issue in ultrasound

imaging. This can be divided into the following:
• Image resolution (see Chapters 3 and 5).

• Perceived appearance – image smoothness, contrast, etc.
• The presence of artefacts within the image.
Resolution is the ability to image fine detail spatially or temporally (fast events).
Apart from the intrinsic resolution that a particular beam and pulse shape can give,
related to frequency and focusing, a number of other factors affect the resolution
seen. The intrinsic resolution is the best resolution achievable from a given trans-
ducer with a chosen set up of frequency, focus, and processing. Other factors such
as tissue type and patient movement will tend to degrade what is actually resolved
in the image.
ACOUSTIC WINDOW
The position of the probe on the skin and the angle of approach to the target of
interest defines what is called the acoustic window. The choice of acoustic window
can make a very big difference to the quality of the image seen, e.g. avoiding excessive
overlying fatty tissue or gas filled gut when observing the abdomen. Some approaches

111
112 Chapter 7 Image Quality and Artefacts
to a target provide a particularly good acoustic window, e.g. using a filled bladder as
an acoustic window when observing the pelvic region.
FRAME RATE: FRAMES PER SECOND (fps)
Frame rate is the rate at which the ultrasound image information is refreshed. The
display shown on the screen is refreshed many times a second so as to display a contin-
uous image. However, depending on a number of factors, the ultrasound information
being displayed may not be acquired fast enough so as to show no flicker or ‘joltiness’
when imaging very fast moving structures such as the heart valve leaflets.
The rate at which new images can be acquired determines the temporal resolution –
the separation of events in time.
The human eye has a temporal resolution of ~40 ms which equates to 25 fps
(Recall: frequency = 1/time)
An image with <20 fps will show ‘image flicker’.
What Determines Frame Rate?
• The number of ultrasound lines/beams in a frame

• The depth of image. The next pulse cannot be sent until echoes from the last
pulse return from the greatest range.
2d 1
Time for one image line t c 1540 ms
c
So, Δt = 13 μs cm−1 (‘go and return’)
100 lines to 10 cm depth 100 130 13000 s so frame rate 76 fps max
We need to leave a gap for echoes die away before the next pulse is transmitted, so
frame rate will always be lower than the maximum possible.
As seen in Figure 7.1, a sector probe will need more lines to fill in gaps distally as
beam directions spread out (although distal line density need be no more than linear
array line density).
Higher line density

Lower line density
FIGURE 7.1 Reduced line density in the far field of a sector scan format.
Interlacing Scan Lines 113
Example: 5.5 cm linear array with focused beamwidth of 0.4 mm needs 137 beams
to cover the width of the array field of view with non-overlapping beams.
A typical linear array has 160 elements.
This could yield a maximum frame rate of ~100 fps but in practice frame rate
will be significantly lower as a result of multi-zone transmit focusing, compound
imaging, etc.
Very much higher frame rates can be achieved using synthetic aperture
techniques (see Chapter 13).
There are a number of ways that frame rate can be improved, which also improve
the appearance of the displayed image.
Where temporal resolution is critical, e.g. in cardiac work, then it is important to
minimise the number of pulses required to obtain an image. The ways in which this
can be done are as follows:
• Reduce the size of the displayed field, e.g. reduce the sector width in a sector
scan. This reduces the number of scan lines required to cover the field.
• Reduce the depth displayed to what is necessary. This reduces the ‘go and
return’ time for each pulse.
• Use a single transmit focus set to the depth of the target of interest. Multiple
transmit zones require a separate transmit pulse for each zone.
In addition to adjusting these parameters, the machine may use interlacing of

scan lines and interpolation to reduce the actual number of transmitted scan lines.
INTERLACING SCAN LINES
The line density in an image can be increased by the use of interlacing (Figure 7.2). To
form an interlaced image frame, the field of view is covered in two sweeps or fields.
In the first field, beams are sent out along the odd numbered lines. In the second
Field 1
Field 2
FIGURE 7.2 Illustration of interlacing with the field 2 angled with respect to field 1.
field, beams are sent out along the even numbered lines. The lines in the second
field may be angled very slightly with respect to those of the first field to ‘fill in
the gaps’. A complete displayed frame is then made up of two interlaced fields
that are both stored and displayed as one image. It has the advantage that each
field only requires half the number of beams, although there will be some blurring
(loss of temporal/spatial resolution) for fast moving structures where movement
has occurred between the two fields being acquired. Because the two fields are
angled slightly to one another, the speckle pattern from each will be different and
the image will appear smoother. A further development of this idea is compound
imaging (see below).
INTERPOLATION – WRITING IN ‘EXTRA LINES’
Frame rate can be improved at the expense of resolution by using interpolation

(Figure 7.3). This involves transmitting a smaller number of scan lines and filling in
the gaps by creating synthetic lines (i.e. no new data). Interpolation has been used to
fill out sector images as the line density reduces distal to the transducer.
The pixels in the inserted lines may be an average of the adjacent real line data
or may use a more sophisticated algorithm taking into account local area brightness,
emphasising edges, etc.
Pros
• Improves frame rate
• Improves image appearance cosmetically
Interpolated lines True scan lines
FIGURE 7.3 Interpolated lines generated between echo generated lines for a linear format and a
sector scan.
Speckle 115
Cons
• The interpolated lines do not contribute new data about the tissue being
examined.
• There is a reduction in resolution within the image as some information from
between the scan lines is missing, even though the ultrasound beams may be
highly focused to be narrow throughout their range.
SPECKLE
Speckle results from the fact that ultrasound imaging uses coherent waves – it is a
coherent imaging system. This means that the phases of the sound waves that form
the ultrasound pulses are all aligned such that peaks and troughs of the waves can
interact through superposition to create maxima and minima in the ultrasound beam
and produce speckle in the echoes (see Chapter 2).
Because the speckle pattern depends on the microscopic positioning of reflectors
smaller than the wavelength of the ultrasound, it is a random noise on the image
where the peaks and troughs appear in different positions each time the tissue moves
or is interrogated by an ultrasound beam from a different direction.
Question: Is Speckle an Artefact?
• It is the way tissue is seen on the image.

• It is a random speckle pattern, not an image of microscopic structure.
• Different tissues will have different speckle patterns depending on their micro-
scopic structure, e.g. normal liver versus metastases, so the speckle pattern
seen does give information on the underlying nature of the tissue.
• Speckle can also be considered to be noise against which we want to see truly
resolved targets.
A number of techniques are used to reduce speckle. These are as follows:
• Frame averaging
• Compound imaging
• Adaptive filtering techniques applying more complex algorithms to
smooth speckle
The ideal would be the following:
• To see edges of organs and structures large enough to be resolved including

changes in tissue parenchyma, e.g. metastases
• To smooth the image where targets are too small to be resolved
FRAME AVERAGING OR PERSISTENCE REDUCES SPECKLE

IN THE IMAGE (‘GRAININESS’)
In order to smooth over the random speckle seen within each frame, a percentage
of the previous frame values are added to the current frame on a pixel by pixel basis
(Figure 7.4). For example, 20% of each pixel value from the last frame is added to the
present frame. The speckle will change with the minute movements of the transducer
from frame to frame and so average out, whereas real structures in the image will
not change (Figure 7.5). The use of a proportion of previous frames is clearly seen in
Figure 7.27, where the image of each of the frames is spread out in time.
+ =
20% of N–1 100% of N
True target
image
FIGURE 7.4 Diagram showing the principle of frame averaging to reduce the effect of speckle.
(a) (b)
FIGURE 7.5 Showing the effect of frame averaging. No frame averaging in (a) and averaging six
frames (b). Note how the speckle in the image is smoothed.
Spatial Compound Imaging 117
USER CONTROL
The degree of frame averaging may be selected by the user. The more frame averaging
used the greater will be the loss of temporal resolution for moving targets. If a lot
of frame averaging is used (high percentage over several frames), then there will be
some noticeable blurring of the image as the probe is moved on the patient’s skin.
It is important to ensure the target image is as stationary as possible before freezing
the image.
In addition to the speckle pattern not directly displaying microscopic information
about the tissue, there is frequently a more general degredation of image quality
caused by a small amount of image data appearing in the wrong place within an
image. This goes under the generic name of clutter.
SPATIAL COMPOUND IMAGING
Curved surfaces are poorly visualised as the surface curves towards the direction of
propagation.
Using beam steering, compound imaging sends multiple beams (three to nine
beams) at different angles into the scan plane so more edges are perpendicular to the
beam (Figure 7.6). This gives better visualisation of curved surfaces. It also acts to
smooth speckle and clutter in the image.
NOTES
• The frame rate is not greatly affected. Acquisition of the images from each
direction is in the form 1–2–3, 1–2–3, for example. The scanner stores the
image from each transmitted direction and combines it with the images
from the other directions dropping the oldest and retaining the newest with
each field sweep.
• The image displayed is the sum of the echoes from beams from multiple
directions.
• Compound imaging also smooths out speckle as it averages the speckle
from multiple beams cf. frame averaging.
• For the same reason, it also reduces clutter.
• There may be some loss of resolution if the speed of sound along the different
directions of the beams is not uniform, leading to some blurring of the
image (see section ‘ARTEFACTS’).
• Activating compound imaging is a user control.
Clearly seen
Poorly seen
FIGURE 7.6 The improvement in visualisation of curved surfaces gained by using

compound imaging.
ADAPTIVE FILTERING
Adaptive filtering and other advanced image processing algorithms aim to enhance
real structures in the tissue and reduce speckle and noise. These algorithms may
include edge enhancement.
Edge Enhancement
Edge enhancement aims to make real edges in the field of view visible and contin-
uous on the image whilst suppressing spurious echoes that may look like edges.
As an example, one such adaptive filter performs an analysis of the image in one
step and then then applies the filtering in a second step (Figure 7.7). The filtering
is called ‘adaptive’ because it adapts what it does depending on what it finds in the
analysis stage. It endeavours to identify real structures both within an image and
across images from frame to frame. As well as trying to detect real structures and arte-
facts, it uses texture properties and statistical analysis of the images at several scales.
In the enhancement phase, it applies edge enhancement for interfaces and smoothing
of uniform areas to suppress speckle and noise [1].
Such filters may be activated as a user control.
CONTRAST RESOLUTION
This is an important concept that describes the ability to detect one target against
another, e.g. the ability to detect a subtle change due to a tumour in normal tissue.
Contrast resolution is partly subjective and depends on the following:
• The size of the target

• The contrast difference between the target and the surrounding image
• How ‘noisy’ the image is, including speckle, clutter, and poor visualisation
• Greyscale mapping and dynamic range used
(a) (b)
FIGURE 7.7 Showing the effect of applying one example of an adaptive filter. Filter off (a) and
filter on (b). Note the change in the appearance of the speckle pattern as shown below each image.
• Human visual perception

• Viewing screen set-up
• Background lighting in the room
Noise in an Image
Noise in an image will reduce the ability to visualise and detect targets of interest.
Sources of noise include the following:
• Speckle
• Poor visualisation due to weak echoes at depth. At the limits of penetration, the
electrical noise in the system has a similar amplitude to the echo signal, as seen
when the gain is set high.
• Poor visualisation due to beam distortion from overlying tissue, e.g. some types
of fatty tissue.
• There will be clutter in the image if echo information from side lobes, grating
lobes and multiple reflections is present.
• Probe movement at the moment of freezing an image will degrade the image
and hence detectability of targets.
The greyscale mapping, dynamic range, and gain chosen will also affect contrast
resolution. In particular, the choice of mapping and dynamic range can make the
image more contrasty or smoother in its display of grey levels. This was discussed in
Chapter 6.
Tissue mimicking phantoms that have calibrated changes in greyscale targets
may be used to demonstrate that no subtle changes are being missed in a particular
mapping setting.
Colour Enhancement of Greyscale Images

In addition to choosing a greyscale mapping to use, the user may be able to ‘colour up’
the image. A single colour ‘greyscale’ is used instead of a black and white greyscale.
In particular, some scanners allow a colour enhancement to be used that matches the
visual sensitivity of the human eye within the optical spectrum (Figure 7.8). The eye
is most sensitive around a yellow–green colour.
Human Visual Perception

• The human eye is better at detecting changes of brightness when the change is
across a sharp boundary than when the change is gradual.
• Smaller targets require a bigger change of contrast to be seen than larger targets.
• The eye detects horizontals and verticals better than oblique changes of
brightness.
• With sharp bright illumination, we can detect grey levels that differ by 1%.
FIGURE 7.8 Example of a Morton neuroma showing the effect of colouring the B-mode
image to match the maximum sensitivity of the eye. Source: Sofka et al. [2]/with permission of
John Wiley & Sons.
NOTE
In relation to perceiving target size against speckle, viewing distance and
magnification (zoom) will affect detectability.
How to Set Up a Monitor

Before attempting to view and interpret clinical images, the monitor on which
they are viewed should be properly set up for optimal viewing. For a greyscale image,
the monitor may be set up using the greyscale wedge display usually displayed
at the side of B-mode images. The monitor brightness and contrast should be adjusted
so that peak white shows as peak white and black has no illuminance at all.
Every step in the wedge should be visible as a step. In particular, levels near the black
and white ends should be distinguishable. A starting point is to set the contrast high
and the brightness low and then increase the brightness until all the steps are seen.
In order to see some of the factors that affect viewing on a monitor, there are
many web sites that can be viewed.
Image Flicker
Image flicker on the monitor itself is not really a problem most of the time, as the
displayed image can be re-read to the display from image memory at least 25+ times/s.
However, a low frame rate will lead to a jumpy image as the transducer is moved on
the patient’s skin. It may be also be seen as image blurring as the probe is moved, e.g.
too much frame averaging (persistence).
Ambient Lighting
It is important to recognise the effect ambient lighting in the scan room has on the
ability to see fine detail in displayed images. The position of the lighting may also be
important, e.g. if a light source reflects directly from the viewing screen surface or
off glasses you are wearing, causing flare, the contrast resolution seen in the image
is reduced.
There is an eye–brain adaption to the average brightness of a scene being viewed.
For example, the eye–brain adjusts sensitivity when going from bright sun into a dark-
ened room. This dark adaption takes time as there is a physical adaption (the iris)
and a biochemical-neural adaption. To become fully dark adapted on a dark night
when looking at stars takes about 20–30 minutes.
The contrast between the average screen brightness and the level of ambient
lighting should therefore be considered when scanning.
ARTEFACTS
DEFINITION
An artefact is any feature in an image that is not an exact or accurate one-to-one
representation of the target being imaged.
Artefacts need to be clearly recognised and understood by the sonographer so as to

avoid mis-interpretation of the images. Some artefacts can be useful.
Assumptions
Ultrasound imaging is based on the following assumptions:
1. Speed of sound is constant in the body.

2. Attenuation in tissue is constant and uniform throughout the image.
3. The beam axis is straight throughout the range of the beam.
4. Later echoes result from targets at greater distance from the probe.
5. The ultrasound beam is infinitely thin, i.e. no echoes arrive that are away from
the beam axis.
6. The image is acquired instantaneously.
Artefacts will occur when these assumptions are not met. They can be considered
around these basic assumptions.
SPEED OF SOUND ARTEFACTS
The assumption is that the speed of sound, c = 1540 ms−1, but it does vary between soft
tissues by ±5%
Refraction
Refraction leads to distortion of the image (see Chapter 2)
Refraction results from changes in the speed of sound at interfaces where the
beam is not perpendicular to the interface. The machine does not know the beam
has changed direction and distal targets will be misplaced on the image, distorting
the structures seen (Figure 7.9). This artefact involves assumptions 1 and 3 men-
tioned above.
Speed of sound differences with in the tissues will cause some degradation of
the image when compound scanning is being used. The effect of this is apparent in
Line of beam
Beam assumed
by scanner
Refracted beam
FIGURE 7.9 Diagram showing how refraction causes distortion in the image by misplacing
targets, e.g. from A to B.
a scan of a phantom with an inclusion having a different speed of sound as seen in

Figure 7.10. Note the appearance of the wires in the image. However, such degrada-
tion will be generally the case across the whole clinical image. Figure 13.12 shows the
effect of correcting for speed of sound on this phantom.
Examples:
• Poor visualisation of deep tissue due to distortion of the beam by overlying fatty
tissue (Figure 7.11) (see Chapter 2).
• A biopsy needle may appear bent as it crosses a tissue boundary where the
speed of sound changes, like a pencil appearing to be bent when put in a glass
of water, where the light moves more slowly in water than in air causing refrac-
tion of the view of the pencil (Figure 7.12).
• Lensing may be seen in a cardiac scan where a dense structure causes the ultra-
sound to be refracted to give a double image of a reflecting target (Figure 7.13).
Axial Misplacement
Axial misplacement (caused by assumptions 1 and 4 not being the case)
The example shown in Figure 7.14 shows the error in measuring the diameter of
a blood vessel due to axial misplacement.
Blood vessel In blood c = 1570 ms−1

Machine assumes c = 1540 ms−1
Difference 1.9%
So, A–P diameter has ~2% error in measurement.

0 0
5 5
10 10
z (mm)
z (mm)
15 15
20 20
25 25
30 30
35 35
40 40
0 10 20 30 0 10 20 30
(a) x (mm) (b) x (mm)
0 0
5 5
10 10
z (mm)
z (mm)
15 15
20 20
25 25
30 30
35 35
40 40
0 10 20 30 0 10 20 30
x (mm) x (mm)
(c) (d)
FIGURE 7.10 Tissue mimicking phantom with (a) a uniform speed of sound (SoS) of 1520 ms−1
imaged from a single forward angle and (b) with compounding. A phantom with an inclusion
having a different SoS (1470 ms−1) imaged from a single forward angle (c) and with compounding (d).
Source: Jaeger et al. [3]/IOP Publishing/CC BY 3.0.
Axial misplacement may also be seen when the beam passes through a tissue, such
as a lipoma, where the speed of sound is slower than 1540 ms−1. The posterior struc-
tures will then be placed more distally on the image than they should be (Figure 7.15).
As we have seen, differences in speed of sound can become a problem when mul-
tiple beams are used to image a target from several directions, such as in compound
imaging. If we could use the true speed of sound for each tissue layer, image quality
could be considerably improved. A first-order correction to the speed of sound
problem is to choose a different speed of sound for the scanner to assume for the
system velocity and see if that improves resolution in the image. In other words,
to use a different average speed of sound than 1540 ms−1. This can improve image
quality as the ultrasound paths from different elements towards a focus will be better
matched to the average speed of sound thereby improving the focus and hence image
(a) (b)
FIGURE 7.11 Poor visualisation due to distortion of the ultrasound beam by overlying tissue
(a) is similar to the distortion of an image caused by ‘wobbly’ glass (b). Source: Oates and
Taylor [4]/with permission of SAGE Publications.
(a) (b)
FIGURE 7.12 Image of a needle in tissue appearing to be bent due to refraction occurring (a).
Source: Courtesy Tim Lees. (b) Optical refraction for comparison.
B2
B1
Lens
Double
image
Object
(a) (b)
FIGURE 7.13 Lens artefact. B1 is the direct beam. For beam B2, the ultrasound is refracted
towards a target to produce a second image of the target along the line of refraction. The example
shows a double image of the aorta (full arrowhead) in a subcostal short-axis image of the heart,
due to refraction of the ultrasound beam at perihepatic fatty tissue (red arrow). A Swan–Ganz
catheter in the right ventricular outflow tract is doubled as well (empty arrowhead). Source:
Bertrand et al. [5]/with permission of Elsevier.
Tissue
1540 ms–1
Blood
1570 ms–1
Image of
posterior wall
FIGURE 7.14 Diagram showing the misplacement of the distal wall of a blood vessel due to
speed of sound error when the machine assumes c = 1540 ms−1.
quality. One technique to achieve this is to try a different speed of sound and examine
the image to see if edges are more sharply defined. A number of speeds may be tried
to find the best solution and different speeds may be used in different regions of
the image. This technique to improve speed of sound correction has been used and
automated in some scanners [6].
Full speed of sound correction is discussed in Chapter 13.
Lipoma 1450 ms–1
Image of
posterior layer
FIGURE 7.15 Diagram showing axial misplacement of distal tissue due to speed of sound in
lipoma being less than 1540 ms−1.
ATTENUATION ARTEFACTS
The scanner assumes a constant attenuation coefficient such that when TGC is cor-
rectly adjusted, the image has uniform brightness throughout the field of view.
Poorly Adjusted Time Gain Control (TGC)

Poorly adjusted time gain control (TGC) can cause information to be missed due
to insufficient gain or due to excessive gain (see Figure 6.4).
Acoustic Shadowing
Acoustic shadowing is caused by structures that are strongly reflective or have
higher attenuation than surrounding soft tissue. It is the acoustic analogue of optical
shadowing such as would be caused by a solid object blocking the beam from a torch.
Distal tissues are not insonated and appear dark on the image.
Acoustic shadowing may be caused by the following:
• Hyper-echogenic targets, e.g. bone, gas, calcification reflecting or absorbing

nearly all the sound impinging upon them (Figure 7.16).
• Oblique surfaces of targets, e.g. arterial wall in cross section, where
the beam is deflected away from distal targets so they are not insonated
(Figure 7.17).
NOTES
• Shadowing restricts viewing angles, e.g. ribs and bowel gas, when distal
tissues need to be seen.
• Shadowing may be useful as it shows a solid acoustically opaque target.
Shadow
(a) (b)
FIGURE 7.16 Shadowing caused by an acoustically opaque target, in this case a rib.
Shadow
(a) (b)
FIGURE 7.17 Transverse view of the right neck showing shadowing due to the ultrasound beam
hitting the common carotid artery wall at an oblique angle and being deflected into adjacent tissue.
Post Cystic Enhancement

Post cystic enhancement is caused by the attenuation being lower than expected
(Figure 7.18).
Beam A in solid tissue has TGC applied to give a uniform image.
Beam B passes through non-attenuating fluid. TGC increases the gain of the
echoes throughout the range of the beam, but as there is almost no attenuation in the
fluid, distal targets have a stronger pulse hitting them. Their stronger echoes are then
amplified too much and show up brighter than adjacent targets from beam A, hence
post cystic enhancement
A B
Fluid
Post cystic
enhancement
(a) (b)
FIGURE 7.18 Post cystic enhancement caused by the passage of ultrasound through a low
attenuating fluid, in this case the bladder.
NOTE
This is a useful artefact – it tells you the structure is fluid filled and is not solid.
Shadowing from the Transducer Surface

Shadowing from the transducer surface results from poor contact with body, e.g.
lack of gel on probe/skin. This can be identified as a shadow that extends throughout
the image from the probe-skin interface (Figure 7.19).
Loss of
skin contact
(a) (b)
FIGURE 7.19 Absent echoes from the transducer surface due to poor acoustic contact at the skin.
REFLECTION ARTEFACTS
Reflection artefacts result from assumptions 3 and 4 above, not always being the case.
Mirror Image Artefact

Mirror image artefact may occur where there is a plane reflecting surface (Figure 7.20)
The scanner assumes the beam is straight – echoes from R are therefore shown at T
on the image.
Examples of reflection artefact may be seen in the lung when the probe is angled
cephalad towards the diaphragm (Figure 7.21). Liver echoes may be seen in the lung, a
gas filled space that should have no echoes within. In a second example, the subclavian
artery may be seen reflected in the upper surface of the lung (see Figure 9.20).
FIGURE 7.20 Diagram showing how reflection at an interface can cause targets R to be
misplaced on the image at T to give reflection artefact.
Liver
Lung
(a) (b)
FIGURE 7.21 Cephalad angulation onto the diaphragm through liver. Liver echoes are seen in
the lung space.
Reverberation
Multiple reflections between parallel surfaces (these artefacts involve assumption 4).
Reverberations can show themselves in several forms creating a number of named
artefacts.
Reverberation Artefact
Reverberation artefact may often be seen when two or more interfaces lie parallel
to one another (Figure 7.22). When the beam is perpendicular to these surfaces, the
pulse can reverberate between these surfaces producing a string of separate echoes
back to the probe. As these arrive at a later time than the first echo, they are
positioned on the image at greater depth. For two parallel surfaces, the second surface
is reproduced at the same depth below the first surface as the distance between the
two surfaces, since the pulse has travelled the same distance a second time. The rever-
beration echoes get progressively fainter as the sound is attenuated along its multiple
echo path. An example is shown in Figure 7.23.
Probe – Interface Reverberation

Reverberation between a bright reflecting target and the probe surface gives a second
echo at twice the depth of the target (2 × d), where d is the target range. This is easily
detected – move the probe up and down on the skin and the reverberation echo will
move twice as fast as adjacent true echoes (Figure 7.24).
Ringdown
Ringdown is caused by reverberation within the front face of the probe, i.e. within
the lens and matching layer, and appears as a thin region of bright echoes immediately
adjacent to the probe-skin surface on the image (Figure 7.25). With the use of good
matching layers and low impedance materials in the probe face, this artefact is
greatly reduced.
t1 t2 t3 t1 Target
t2 Reverberation
t3 echoes
FIGURE 7.22 Diagram showing how reverberation artefact is generated by multiple reflections
from parallel surfaces.
Reverberation
Reverberation
echoes
(a) (b)
FIGURE 7.23 Reverberation seen within the lumen of the carotid artery arising from multiple
reflections at the proximal wall.
Comet Tail or Ringing

There are two separate causes that give a similar appearance on the image.
• Multiple reflections within a solid target. The speed of sound within the target
is high and the size of the target is such that the multiple echoes do not show
up as individual echoes but merge to produce a streak or train of echoes in the
image (Figure 7.26).
• Small air bubbles or fluid trapped between air bubbles may be caused to
resonate when hit by the ultrasound pulse. They then go on ringing as a
bell would when hit, producing a string of echoes that appear as a streak in
the image.
NOTES
• Comet tail is useful for detecting foreign bodies in tissue, e.g. IUCD and
wires/staples.
• Small bubbles of air show as bright echoes with streaks in the image. This
may be useful, e.g. to see where a needle introducing fluid is placed. It can
also be a problem when it blocks viewing distal tissues.
Move probe
up and down
on skin
Reverberation
Reverberation echo
moves twice as fast
against background
Tissue moves at
same speed as
transducer
FIGURE 7.24 Reverberation artefact arising from reflection at the transducer surface and how to
detect its presence by moving the transducer up and down on the skin surface.
Ringdown
(a) (b)
FIGURE 7.25 Ringdown from the front face of the transducer seen with the transducer in air.
When in contact with the skin this ringdown will be less apparent. Source: Courtesy of Barry Ward.
Train of echoes
behind target
(a)
(b)
FIGURE 7.26 Comet tail artefact seen behind a biopsy needle. Source: Courtesy of Tim Lees.
ANISOTROPY
Anisotropy describes the case when a tissue changes its appearance, scattering sound
differently, as the orientation of the target with respect to the scan plane changes. The
tissue is then said to be anisotropic.
Some tissues such as muscles and tendons have a filamental structure with
strongly oriented cells or fibres. These can act in concert together to alter the echo-
genicity of the tissue depending on the angle of insonation (Figure 7.27). Changing
the orientation of the beam from along to across the axis of the tissue fibres may
change the appearance of the tissue from a hyperechogenic tissue to a hypoechogenic
tissue. A change of just 5° in angulation of the beam may be sufficient to produce this
change. This is a reflection artefact as the reflectivity of the tissue changes with orien-
tation of the ultrasound beam.
Beam Shape Artefacts 135
(a)
(b)
FIGURE 7.27 Anisotropy seen in a superficial muscle imaged (a) longitudinally and (b)
transversely. Note the different speckle pattern texture seen in each case.
NOTES
• Mis-diagnosis is possible, e.g. a normal tendon may appear damaged or ruptured.
• Tendons may be differentiated from nerves, which do not show anisotropy.
• Examine potential anisotropic targets from more than one direction.
• Compound imaging will reduce the effect of anisotropy as the image is
formed using beams from several directions.
BEAM SHAPE ARTEFACTS
Beam Shape Artefacts – caused by assumption 5 not being the case
Slice Thickness Artefact

Slice thickness artefact is caused by inclusion of targets outside the image plane
but within the elevation beamwidth. This results from the fact that the beam and scan
plane is not infinitely thin. An example from within an artery is shown in Figure 7.28.
ry
A Arte
B
C
D
A B
C D
Image plane
(a)
(b)
FIGURE 7.28 Slice thickness artefact seen within the brachial artery imaged in longitudinal
section. The vessel wall outside the scan plane is still within the slice thickness of the beam.
Side Lobes and Grating Lobes

Side lobes and grating lobes: These generally degrade the detail in the image as
echoes come from other directions than the beam axis and are added to the beam
image. They may be seen in cystic structures such as the bladder where diffuse echoes
may be seen within the cystic fluid. They are also specifically seen in echocardio-
graphic scans where a strong reflector is seen stretched out into a space beyond its real
position (Figure 7.29).
Slice thickness artefact, grating lobe, and side lobe artefact will be present to some
degree in all images. They contribute to ‘clutter’ in the image and generally degrade
the image seen. However, it is important to recognise them as artefacts in specific
situations where they are seen, such as those indicated in the examples given above.
Temporal Artefacts 137
Object
(a)
(b)
FIGURE 7.29 Echoes from the side lobe register along the main beam direction, then the main
beam and opposite side lobe in turn pick up echoes to give a streaked appearance for the object.
Example shows cardiac parasternal long axis view with linear side lobe artifact (arrow) in the
ascending aorta due to a calcified sinotubular junction (arrowhead). Source: Bertrand et al. [5]/
with permission of Elsevier.
TEMPORAL ARTEFACTS
Result from the fact that the image is not obtained instantaneously (assumption 6).
If the frame rate is low then fast moving targets will not be adequately tracked,
they will move in the time it takes to acquire one frame. Artefacts will result, as may
be shown using a rotating wire phantom (Figure 7.30a). A single straight line of wires
crosses the image plane and when stationary appears as in Figure 7.30b. When the
phantom is rotated, the image appears as in diagram Figure 7.30c. In the image shown
Double image
from frame averaging
Double image
from interlacing
Curved image
from time distortion
FIGURE 7.30 Diagram of a moving phantom to show temporal artefact.
FIGURE 7.31 An image taken from the moving phantom shown in Figure 7.30. Duplication
giving image pairs is due to interlacing of fields. Successive pairs are due to frame averaging. Each
line of echoes is curved due to movement over the time taken to acquire the image.
in Figure 7.31, there are three sets of double images. The pairing is caused by inter-
lacing of two image fields and each line of wires is not straight but is distorted due
to movement of the target during the time the frame was acquired. The three sets of
images are the result of frame averaging. Each set is weaker than the previous set as
the proportion of its contribution to the image reduces.
Final Example 139
FIGURE 7.32 Image of a carotid artery showing the effect of movement of the probe during
moment of freezing the image causing blurring. This was exacerbated by the use of multiple
transmit focal zones and frame averaging.
Image Blurring
Use of high levels of frame averaging can cause image blurring on the live image dur-
ing movement of the probe on the skin.
Image blurring on frozen images may be caused by movement of the probe
at the moment of freezing the image (Figure 7.32). This may be due to hand
movement by the operator or movement in the subject, e.g. patient breathing or
foetal movement. This will cause loss of detail in the image and possible errors in
making measurements from the image.
FINAL EXAMPLE
The final example of the need to interpret images in the presence of artefacts is a
reminder that it is also important to know the anatomy being viewed. Figure 7.33
shows a transverse view of the lower neck. Within the larynx there are a lot of echoes.
We know that the larynx is an air-filled structure and that ultrasound will not pene-
trate air. Therefore, all of the echoes in the normal larynx must be artefactual. This
image also shows a number of other examples of the artefacts discussed.
FIGURE 7.33 Transverse view of the lower neck. All of the echoes seen within this normal
larynx must be artefactual as the larynx is air-filled and ultrasound does not penetrate an air
interface.
REFERENCES
1. Meuwly, J., Thiran, J., and Gudinchet, F. (2003). Application of adaptive image processing tech-
nique to real-time spatial compound ultrasound imaging improves image quality. Investigative
Radiology 38: 257–262.
2. Sofka, C.M., Lin, D., and Adler, R.S. (2005). Advantages of color B-mode imaging with contrast
optimization in sonography of low-contrast musculoskeletal lesions and structures in the foot
and ankle. Journal of Ultrasound in Medicine 24: 215–218.
3. Jaeger, M., Robinson, E., Akarçay, H.G. et al. (2015). Full correction for spatially distributed
speed-of-sound in echo ultrasound based on measuring aberration delays via transmit beam
steering. Physics in Medicine and Biology 60: 4497–4515.
4. Oates, C.P. and Taylor, P. (2016). Helping expectant mothers understand inadequate ultrasound
images. Ultrasound 24: 142–146.
5. Bertrand, P.B., Levine, R.A., Isselbacher, M.E. et al. (2016). Fact or artifact in two-dimensional
echocardiography: avoiding misdiagnosis and missed diagnosis. Journal of the American Society
of Echocardiography 29: 381–391.
6. Napolitano, D., Chou, C., McLaughlin, G. et al. (2007). Sound speed correction in ultrasound
imaging. Ultrasonics: Supplement 44 (Suppl 1): e43–e46. https://doi.org/10.1016/j.ultras.
2006.06.061.
CHAPTER 8
Principles of Doppler
Ultrasound
THE DOPPLER EFFECT
The Doppler effect is the change in the detected frequency of a sine wave when there
is relative motion between the transmitter and the receiver. This will occur for all
wave phenomenon. It is the change in pitch you hear when a motor bike races past you.
The pitch is higher as it comes towards you and drops to a lower pitch as it moves away.
NOTE
For historical reasons and as it is the name of a person, Christian Doppler, who
first described the effect, ‘Doppler’ is always spelt with a capital ‘D’.
To see why there is this change in received frequency, consider the example of a boat
rowing towards a source of waves on a lake (Figure 8.1). It will intercept those waves
more rapidly than if the boat was stationary. If it were being rowed away from the source,
the boat would intercept the waves at a lower frequency. If the boat was stationary,
it would bob up and down at the same rate as the waves were being generated.
In this case, the source of waves is stationary and the boat is a moving receiver.
When a fire engine passes you at speed, with its sirens sounding, you hear a
change in the pitch or frequency of the sound. The frequency is higher as it comes

141
142 Chapter 8 Principles of Doppler Ultrasound
FIGURE 8.1 Illustration of a boat moving at velocity ‘v’ encountering wind-blown waves moving
at velocity ‘c’.
towards you and lower as it moves away. In this case, the fire engine is a moving
transmitter and you are a stationary receiver.
It doesn’t matter which is moving and which is stationary; it is the relative
motion between the transmitter and receiver that causes the Doppler effect.
In the case of clinical ultrasound, the Doppler effect will occur when there is
relative motion between an echo target and the transducer, e.g. moving blood or
cardiac wall. We can use this phenomenon to non-invasively detect movement and
measure velocities of targets within the body.
The change in frequency between the transmitted and received sound waves at
the transducer is known as the Doppler Frequency or Doppler Shift.
Imagine a little man on a red blood cell (RBC), the ultrasound target, that is
moving directly towards the transducer at a velocity, v (Figure 8.2). The transducer
(transmitter) is stationary and he is moving. He will see an increase in frequency
fT fT fT as he intercepts the sound waves due to his motion towards the trans-
ducer. He then re-transmits the higher frequency sound waves fT as an echo back to
the transducer. We now have the situation of a moving transmitter (the RBC man) and
a stationary receiver (the transducer).
As the RBC man moves towards the transducer, the distance between successive
wave crests λR in his reflected wave will be shorter. A shorter wavelength λ implies a
higher frequency (c = f·λ). So again, because of their relative motion, the transducer
will see an increase in frequency in the echoes, on top of the already increased fre-
quency, reflected back by the RBC man fR fT fT .
So, in the course of the ultrasound being sent out from the transducer and the echo
coming back, there have been two Doppler shifts, one on the way out and one on the
way back, to produce the total Doppler shift detected by the transducer fR = fT + 2ΔfT.
This gives the factor of ‘2’ in the Doppler Equation (shown below). If the RBC man
was moving away from the transducer the Doppler effect would cause a lower fre-
quency to be detected, again with a shift on the way out to the target and another on
the return path and the sign would be negative, i.e. fR = fT − 2ΔfT.
The Doppler Equation 143
C
RBC “sees”
(c+ v)
fT′ =
λT
v
c
λ=
fT
C
Backscattered
frequency fT′
v
(c–v)
λR =
f′ T
FIGURE 8.2 Illustration of the changes in frequency and wavelength seen on the outward path
and the return path of ultrasound reflecting from moving blood.
THE DOPPLER EQUATION
The Doppler equation gives the change in frequency detected by a transducer when
ultrasound has reflected off a moving target in the body. It assumes that the target
velocity (typically, 0–10 ms−1) is much smaller than the speed of sound in tissue
(1540 ms−1), which is true.
v cfD
fR fT fD 2 fT cos D
so v cos D
c 2 fT
where,
fR is received frequency
fT is transmitted frequency
fD is the Doppler frequency or Doppler Shift
v is the target velocity
c is the speed of sound in blood
The term cos(θD) is needed as only the velocity directly towards or away from the
transducer generates a Doppler shift. The example of the man on a RBC assumed that
he was moving directly toward or away from the transducer. In general, the target
might be moving at some angle θD to the line of the ultrasound beam. The compo-
nent of velocity toward or away from the transducer is then given by cos(θD) and θD is
known as the Doppler angle (Figure 8.3).
θD
v
vz
θD
vz = v.cos θD
FIGURE 8.3 Definition of the Doppler angle θD and the component of velocity vz toward the
transducer.
NOTES
• cos 0° = 1 The Doppler shift will be a maximum for a target moving directly
along the line of the ultrasound beam toward or away from the transducer.
• cos 90° = 0 If the target is moving perpendicular to the ultrasound beam
there will be no Doppler shift.
• cos 60° = 0.5 -easy to remember.
• For most echocardiography work a line of sight along the direction of
motion can be obtained and so θD = 0 and cos(θD) = 1. So, for cardiac work,
the Doppler equation can be simplified to leave out the cos(θD) term.
cfD
The Doppler equation for cardiac work v
2 fT
NOTE
Velocity v is positive when it is toward the transducer and negative when it is away
from the transducer. The change in frequency is then calculated correctly.
DUPLEX ULTRASOUND
Doppler ultrasound gives us a means of making non-invasive measurements of

velocity within the body. A Duplex ultrasound display is one which shows both the
B-mode image and the Doppler Waveform. Superimposed on the B-mode display is a
line showing the Doppler ultrasound beam within the field of view and the sample
CW Doppler 145
Sample volume
Angle cursor
FIGURE 8.4 Diagram and image showing the alignment of the angle cursor and sample volume
placement relative to the ultrasound beam and the vessel being examined.
volume from which the Doppler signal is obtained (Figure 8.4). Within the sample
volume there is a rotating angle cursor that may be aligned with the direction of flow
in the target.
When the angle cursor is correctly aligned with the direction of flow, the scanner
then knows the value of θD. The scanner is then able to solve the Doppler equation
to find the absolute velocity of the target; since it knows the transmit frequency and
the speed of sound, we tell it the Doppler angle from the B-mode image, the machine
compares the received frequency with the transmitted frequency to find the Doppler
frequency. The target velocity v is thus found non-invasively.
EXAMPLE:
fT = 6 MHz
c = 1540 ms−1
v = 0.6 ms−1
θD = 60° then cos θD = 0.5 and fD = 2.4 kHz which is in the audible range
Since fD is in the audible range, the simplest way to detect the Doppler frequency
is to amplify it and put it through a loudspeaker. This is the sound you hear when
performing a Doppler examination.
CW DOPPLER
In a continuous wave (CW) Doppler system, ultrasound is continually being

transmitted by one transducer and continuously being received by another, e.g. hand
held Doppler used to detect peripheral pulses, or foetal heart monitor.
Figure 8.5 shows how the transmit and receive beams are arranged so that they
overlap over a long thin line. The sensitive volume within which moving targets can
be detected is the cross-over region of these two ultrasound beams when the target
will be ‘visible’ to both beams. The cross-over region is called the Sample Volume.
In general, there will be a range of Doppler frequencies from many targets having
different velocities within the blood stream. Higher Doppler frequencies are produced
by targets moving toward the transducer and lower frequencies are produced by tar-
gets moving away from the transducer.
NOTE
When looking at blood flow, flow in the normal physiological direction is called
Forward Flow and flow in the reverse direction is called Reverse Flow, e.g. in
arteries forward flow is away from the heart and in veins it is towards the heart.
Whether forward flow gives a higher or lower Doppler shift depends on the
direction of flow with respect to the ultrasound beam because the Doppler shift
depends on whether the flow is towards or away from the transducer.
Detection of Direction of Flow

In order to determine the direction of flow we need to separate out the Doppler
frequencies that are less than the transmitted frequency from those that are greater.
The output of the mixer detector effectively strips out the transmitted frequency and
just leaves the detected Doppler frequency. We might expect the reverse flow com-
ponent with lower frequencies to come out of the mixer-filter with frequencies less
than zero as shown on the spectrum Figure 8.7a. However, you cannot have negative
frequencies, a frequency has to have a positive value. In practice what happens is that
the negative frequencies get reflected onto the positive side and forward and reverse
Tx fT oscillator
Tx
Rx
Rx X Audio
Mixer Low pass

filter
FIGURE 8.5 Block diagram of a basic hand held continuous wave Doppler system.
CW Doppler 147
Detecting the Doppler Frequency

The Doppler frequency is detected by multiplying or ‘mixing’ the received signal
with the transmitted signal. This process gives ‘sum and difference’ terms.
i.e.: sum = fR + fT -a very high frequency

and difference = fR − fT = fD -the frequency we want to detect
A low pass filter allows the low frequency Doppler frequency to pass and cuts out
the high frequency sum term.
The waveforms and spectrum show this process (Figure 8.6).
Difference
wave
Low pass
E filter
Sum
wave
fD fT 2fT
(a) (b)
FIGURE 8.6 The sum and difference waves obtained by mixing the echo signal with the
transmit frequency and the use of a low pass filter to recover the Doppler signal fD, shown on
the power spectrum.
flow frequencies overlap and cannot be distinguished. In other words, the very simple
system shown in the block diagram cannot detect the direction of flow.
In order to detect flow direction, we can mix in an offset frequency foff into the
output of our simple system to give the spectrum shown on the right of Figure 8.7. For-
ward and reverse flow are then separated either side of the offset frequency (which is
just a few kilohertz). By filtering each one off as shown, flow direction can be detected
with flow toward the probe sent to one loudspeaker and flow away from the probe sent
to another, for example, left and right in a set of stereo headphones. In practice this
separation is done by a more sophisticated algorithm than simply filtering the signal.
When you hear the sound coming from the loudspeaker on the ultrasound
scanner, you are hearing the actual Doppler shift frequencies being detected.
Low pass High pass
E
R F R F
O O foff
ie: F + R
(a) (b)
FIGURE 8.7 The principle of mixing a detected signal with an offset frequency foff to separate
the forward and reverse flow components. (a) Shows the situation when there is no offset
frequency mixing. Forward and reversed flow signals are superimposed above zero frequency.
(b) By mixing in an offset frequency forward and reverse flow are then shown either side of the
offset frequency.
The higher the pitch, the higher the velocity (assuming a straight vessel) and this
can be used to locate the region of fastest flow in a stenosis -if you have a good ear!
The Doppler Waveform Display (Sonogram)

The technical name for the Doppler waveform display is a sonogram.
We have previously seen examples of waveforms and their frequency spectra, e.g.
the spectra of B-mode pulses. The mathematical algorithm that lets us convert from
the time domain to the frequency domain is called the Fourier transform (cf.
Fourier series, Chapter 5).
In digital processing, a real time signal can quickly be transformed to show
its spectrum using a Fast Fourier Transform (FFT) algorithm to show what
frequencies are contained within the signal.
Examples
Assume that the Doppler sample volume includes the whole cross-sectional area of a
blood vessel.
1. Slowly Moving Blood Flow in a Vessel

The velocity profile (change of velocities across the diameter of the vessel) has
the mathematical shape of a parabola and is known as parabolic flow (Figure 8.8).
The diagram shows the spectrum of the Doppler signal from such slowly moving
blood flow. For parabolic flow, there is a constant volume of blood moving at each
velocity, from zero at the vessel wall up to vmax in the centre of the vessel. This is
reflected in the shape of the power spectrum.
CW Doppler 149
vmax
Blood velocity profile Doppler signal contains

many frequencies
6
5
4
PfD fD 3
PfD
2
fD 1
fmax 0
Grey Frequency
scale bins
Instantaneous
Doppler spectrum
FIGURE 8.8 Diagram showing how a range of velocities gives a broad spectrum of Doppler
frequencies that may be displayed using a greyscale. In this case producing the instantaneous
Doppler power spectrum for slowly moving blood flow.
NOTES
• A spectrum is a display of the strength, or power, of each of the frequencies
present in a signal. For example: the brightness of the colours seen in
a rainbow.
• As the diagram shows the spectrum of frequencies present in the Doppler
signal at one instant in time, it is called an instantaneous Doppler power
spectrum.
• The amplitude of the Doppler signal depends on the number of red blood
cells reflecting back the ultrasound signal, so the amplitude or power P(f0)
of the Doppler spectrum at each frequency is proportional to the volume of
blood moving at each velocity within the sample volume of the ultrasound
transducer.
• The instantaneous Doppler spectrum can be shown as a graph of the power
of the signal vs the Doppler frequency, or it can be shown as a vertical bar
divided into a set of frequency ranges, called frequency bins, with each
level filled in with a grey level that depends on the signal strength in that
range of Doppler frequencies.
2. Fast Moving Blood Flow in a Vessel

The velocity profile becomes blunt and nearly all the blood moves at the same velocity,
known as ‘plug flow’ (Figure 8.9).
6
v
5
4
PfD
fD 3
2
fD
1
fmax
0
FIGURE 8.9 The instantaneous Doppler spectrum for fast flowing blood.
7
6
5
4
fD
3
2 fD signal
1
0
t1 t2 t3 t4 t5 t6 t7 t8 10 ms 10 ms 10 ms 10 ms 10 ms 10 ms
(a) (b)
(c)
FIGURE 8.10 Diagram showing how the sonogram (a) is constructed from a series of
instantaneous Doppler spectra (t1 to t8) each produced from consecutive 10 ms segments of the
Doppler signal (b). Example of a real ultrasound sonogram (c).
As the velocity profile has changed, the corresponding instantaneous Doppler

spectrum will change to show that nearly all of the blood is travelling at a high velocity,
giving high Doppler frequencies. Only a small amount of blood near the vessel wall is
travelling at a low velocity.
From these two examples, we see that in each case the Doppler power spectrum
shows the distribution of frequencies (velocities) of the blood moving through the
sample volume.
The FFT algorithm repeats the spectrum analysis of the Doppler signal very
quickly to produce a series of spectra that form the Doppler waveform display or
sonogram (Figure 8.10).
CW Doppler 151
NOTES
• Each vertical line t1, t2… . is a complete instantaneous Doppler power
spectrum.
• The length of the segment of signal used Ts determines the frequency
1
resolution f e.g. 10 ms segments gives a resolution of 100 Hz. The
Ts
frequency resolution determines how fine are the steps with which we can
display velocities. The significance of this is that if we want to look at low
velocities with some detail, the segment length needs to be longer. If you
think of looking at a very slowly moving object, you have to look at it for a
long time in order to tell that it has moved.
• Each segment (e.g. 10 ms) produces 1 line of spectrum for the sonogram.
• There is a trade-off between the frequency resolution and the rate at which
the number of spectra are produced. For example, 10 ms analysis segments
will enable 100 spectral lines per second with a frequency resolution of
100 Hz to be displayed.
• By mixing in a base-line offset frequency, bi-directional flow can be displayed
(Figure 8.11).
fD 3
2
Offset
1 frequency
0
(a)
(b)
FIGURE 8.11 Showing how the use of an offset frequency enables the display of forward and
reverse flow on a sonogram.
NOTE
The display can be inverted on the screen and it is common to set it so that forward
flow in the physiological direction is shown above the zero line and reverse flow below
the line. That is arterial flow away from the heart, when looking at arteries, and
venous flow toward the heart, when looking at veins, is shown above the zero-flow line.
CW Doppler Summary
• Disadvantage is that any moving target in the sample volume (cross-over
region) will contribute to the detected Doppler signal, that is, it has no range
discrimination. Two blood vessels within the sample volume will both be
detected and their signals superimposed on one another. Where there is only
one main vessel in the sample volume, for example, in the limbs when the
long arteries are being viewed, or the signal being examined is very distinct
for example, when monitoring the foetal heart, this disadvantage is not too
much of a problem. However, in the abdomen where there are many vessels
that might fall in the sample volume, it is a big problem.
• Advantage is that it can detect very high velocities unambiguously. There is
no upper velocity limit (see PW Doppler below). This advantage is useful in
cardiac studies where very fast jets may be seen.
PULSED WAVE DOPPLER (PW DOPPLER) AND RANGE GATING
The way to overcome the CW disadvantage of not being able to isolate specific
vessels so they can be examined individually is to use pulsed wave Doppler with
range gating.
Range Gating is the method of isolating a target at a particular range. It enables
specific Doppler targets to be isolated and interrogated. This is shown in Figure 8.12.
A pulse is sent out and the returning echoes are ignored until echoes from the
target depth arrive. The receive ‘gate’ is then opened and the signal detected. Beyond
the depth of interest, the ‘gate’ is then closed and no more signal is detected. This then
defines the sensitive or sample volume. In other words, we have gated out the signal
just from a particular range -hence range gating.
NOTES
• The sample volume size is a combination of the range gate length Δt and
the pulse length (see below for more on this).
• The user can select the depth of the sample volume and its length. These are
shown on the B-mode image (Figure 8.4).
2Δt
T1 T2
s1
2s
t1= c 1 range gate at s1 ± Δt
FIGURE 8.12 Diagram showing the principle of range gating to isolate the signal from T1 and
reject the signal from T2.
The rate at which samples are obtained is the pulse repetition frequency or
PRF of the pulsed wave Doppler transmission pulses.
Detection of the Doppler Signal

The period of Doppler frequencies (1/fD)is low compared with the go-and-return time
of the pulses. The length of signal gated out from each pulse is too short to determine
the Doppler frequency. Therefore, the Doppler signal we want to detect has to be built
up over many pulses as shown in Figure 8.13.
Tx is the sequence of transmit pulses. The range gate is opened after a delay Δt
from each transmit pulse to allow echoes from the depth of interest to be detected.
Each ‘transmit – gated receive’ sequence produces a sample of the Doppler signal
Tx
Range gate
Δt
Detected signal
Sample + hold
FIGURE 8.13 Diagram showing how the Doppler signal (solid line shown on the detected
signal) is recovered from the range gate depth using a sequence of samples obtained from
successive pulses Tx.
from the Doppler detector (mixer). The detected signal in Figure 8.13 shows the sam-
ples obtained with the true Doppler signal superimposed as a solid line.
The ‘sample and hold’ takes each detected level and holds it until the next value
comes in. This produces the rough shape of the Doppler signal. This is then put
through a low pass filter to smooth it and sent to the FFT analyser in short segments
to calculate the sonogram as previously described.
NOTE
The key thing to note is that in PW Doppler, the Doppler signal is built up from a
sampled signal.
A PW Doppler system is shown in Figure 8.14.
NOTES
• The scale and sweep speed controls affect a number of the components in
the block diagram but are shown on the display for convenience.
• In the case of duplex scanning with an array transducer, the Doppler beam
will have focusing applied as for B-mode with the focus set at the depth of
the sample volume.
Aliasing
In order to detect the frequency of the Doppler signal, the wave must be sampled at
least twice in each period, that is, you need to know that it has at least gone above and
below the zero line in that time. (recall, period T = 1/f).
Tx pulse
Tx generator
Sweep
Depth Baseline speed Scale
fT generator Range gate
Length offset
Spectrum
Rx X Sample + hold analyser
Mixer Low pass
filter
Audio
FIGURE 8.14 A block diagram of the main components of a pulsed Doppler system.
This criteria for sampling is known as the Nyquist Limit. The pulse repetition
frequency PRF of sampling pulses must therefore be at least twice the maximum
Doppler frequency we want to detect.
i.e. PRF 2 fDmax
fDmax is then the maximum Doppler frequency we can unambiguously detect with that
PRF. Fewer samples than this produces a phenomenon called aliasing where the
detected Doppler frequency gives the wrong value.
Aliasing is seen in films when a waggon wheel in a western or an aeroplane propeller
appears to go backwards at certain speeds. What happens there, is that the film is made
of a series of pictures (samples) shown in very quick succession so that your eye per-
ceives continuous motion. However, if between two frames the propeller has gone all
the way round but not quite, and the same again for the next frame and the next frame
etc., then instead of seeing a propeller going round very fast in one direction, you will
see it slowly moving backwards. It is a repeating motion that is being inadequately
sampled (less than the Nyquist limit) and aliasing is occurring.
Imagine watching someone spin a ball on a string and you blink your eyes to get
samples Δt apart to determine which way the ball is spinning (Figure 8.15). This is
what you would see depending on how long Δt is:
(a) A high sample rate -the sample interval is much less than the period of rota-
tion and is less than half the period of rotation. The direction of rotation can
be correctly ascertained as anticlockwise.
(b) The sample rate exactly equals half the period of rotation so you can no longer
tell whether the ball is travelling anticlockwise or clockwise
(c) Sample rate too low -the sample interval is greater than half the period of
rotation but less than the period of rotation. The ball then appears to be going
clockwise, i.e. backwards, due to inadequate or under sampling.
(a)
? ? ? ? ? ? ?
(b)
(c)
Δt
Samples
t0 t1 t2 t3 t4 t5 t6 t7
FIGURE 8.15 Rotating ball on a string showing how inadequate sampling leads to ambiguity
and aliasing in determining the direction of movement of the target.
Tx pulses
Detected
signal
Detected fD True fD
FIGURE 8.16 Diagram to show how inadequate sampling of the Doppler signal leads to the
detection of too low a frequency, an example of aliasing.
The same thing happens when the repeating cycles of the Doppler signal are
inadequately sampled.
In Figure 8.16, the dashed line shows the true Doppler frequency. With the sample
rate shown, the solid line is the frequency that is detected. Due to aliasing, instead of a
high Doppler frequency fDtrue, a low frequency fDaliased is detected.
Appearance of Aliasing on the Sonogram

On the sonogram aliasing appears as the visual ‘wrap around’ of the waveform
displayed. What should be shown as a high frequency ‘wraps around’ to the low fre-
quency end of the spectrum as shown in Figure 8.17.
NOTES
• Aliasing causes the high Doppler frequencies fD to ‘wrap around’ to the
low frequency end of the spectrum.
• In the second diagram (b), aliasing is still seen at the low frequency end of
the sonogram display, but the whole waveform has been shifted up by add-
ing in the baseline offset frequency.
• Aliasing can be avoided by increasing the PRF to give adequate sampling -
the scale control on the scanner increases the PRF to do this.
Maximum Velocity vs Depth
Aliasing affects the maximum velocity we can detect at a given depth in tissue:
2d
Time for echo to return t
c
1 c
So for a given depth d the maxPRF (need to wait for the echo to return
t 2d before sending out the nextpulse)
Nyquist limit 2 f c so c
Dmax
fDmax
2d 4d
v
Doppler frequency fD 2 fT cos D, the Doppler Equation
c c2
vmax
So, maximum velocity detectable without aliasing at depth d is 8d fT cos D
The maximum velocity that can be unambiguously detected, vmax, is reduced at

greater depths as the PRF must be lower to allow for the greater go and return time for
echoes. vmax can be increased by using a lower transmit frequency fT. For this reason,
the transmit frequency of Doppler pulses on a wideband probe is usually less than
Baseline
offset
A A
(a) (b)
(c)
FIGURE 8.17 The appearance of aliasing on the sonogram (a) showing how the high frequencies
‘wrap around’ to the low frequency end of the spectrum. (b) shows the case where there is a
baseline offset and (c) shows an example of aliasing on a real sonogram.
the imaging frequencies used. This allows higher flows to be detected before aliasing
occurs. Increasing vmax without aliasing may be achieved as follows:
• Adjust the PRF using the scale control.

• At any given depth higher velocities can be viewed without aliasing by steering
the Doppler beam toward the normal to the direction of flow thereby reducing
cos θ. This will give some loss in accuracy of velocity measurement (see angle
error below) and at 90° there will be complete loss of signal.
Some scanners have a ‘high PRF mode’. For this they double the PRF which
will double the Nyquist limit for fD at the expense of creating two sample volumes,
one at the target depth and one at half the depth (Figures 8.18 and 12.15). This is not
usually a problem as there is unlikely to be another blood vessel at exactly half the
distance -USER BEWARE!
For cardiac work, where very fast flow jets may be encountered, the
normal imaging transducer may be run in CW mode, thereby avoiding aliasing
altogether.
Observing Low Velocities

We have seen that in order to observe high velocities without aliasing the PRF needs
to be increased by increasing the scale control. In order to observe low velocities
and clearly see these on the sonogram, the PRF needs to be reduced. As discussed in
relation to the sonogram above, if something is moving very slowly then you need
to look at it for a longer time in order to determine that it has moved. Turning the
scale down has the same effect in enabling low velocities to be seen on the Doppler
waveform display.
FIGURE 8.18 Doubling the PRF to double the Nyquist limit to double the Doppler frequency
detectable produces a second sample volume B at the half the range gate depth A.
PW Doppler Summary
• Disadvantage: Subject to aliasing if the Doppler signal is inadequately sam-
pled. There is a maximum velocity which can be unambiguously detected.
• Advantage: Uses range gating to enable individual targets to be isolated for
examination.
PW Doppler Pulses
The pulses used for PW Doppler are generally longer than those used for B-mode
imaging – typically 6–7 cycles long (Figure 8.19). This is because the transmit
frequency needs to be well defined in order accurately detect changes due to
moving targets. By using a longer pulse the spectrum/bandwidth of the pulse
will be a lot narrower but will still not be absolutely precisely defined as shown
in Figure 1.18.
The sample volume length will be defined as a combination of the range gate
length and the pulse length (Figure 8.20). Signals will continue to be detected from
along the length of the pulse as long as the range gate is open. For very short range
gates, the pulse must be made shorter for the signal to be defined as arising from that
range gate.
Doppler signals are noisy signals as shown in the instantaneous power spectrum,
Figure 8.21.
y
x
Narrow
E(f) bandwidth
z
nλ fT f
FIGURE 8.19 The spectrum of a Doppler pulse.
Δt
PL
FIGURE 8.20 The relationship between pulse length PL and the range gate length Δt. Doppler
signals will be detected for the whole time the pulse passes the open range gate.
Ideal Real
E(f)
fD
fmax
FIGURE 8.21 Diagram of the ideal and real instantaneous Doppler power spectrum for
parabolic flow.
NOTES
• The received echo from the sample volume will be modulated by a
speckle pattern arising from the red blood cell reflectors within the
sample volume.
• Each pulse-echo sequence sees a slightly different target because the blood
is moving, therefore the speckle changes with each pulse (a blood target
may be considered stationary over a period of ~10–20 ms).
• As the Doppler sample volume is made shorter, for example to look at a
narrow region within a vessel, the Doppler frequencies become less well
defined. This is due to a phenomena called Intrinsic Spectral Broad-
ening (ISB)
INTRINSIC SPECTRAL BROADENING (ISB)
In a real system, a single target moving at velocity v will produce, not just a
single Doppler frequency f D, as might be expected, but a range of frequencies
f D ± ΔfD. This is called intrinsic spectral broadening or ISB. It results from
the way scanners have to be constructed and run, that is, it is intrinsic to the
real machine rather than what might be expected from an ideal machine and it
cannot be avoided.
There are two causes/explanations for intrinsic spectral broadening [1].
The first cause is that all Doppler pulses will have a finite bandwidth and so
produce a range of velocities across their bandwidth from a single velocity target
(Figure 8.22). The shorter the pulse, the wider the bandwidth and the greater the ISB
will be, i.e. fD ± ΔfD. A very short range gate requires a very short pulse, and hence
there will be greater intrinsic spectral broadening.
Intrinsic Spectral Broadening (ISB) 161
t E(f)
Δt f
fD
FIGURE 8.22 Diagram showing how intrinsic spectral broadening is caused by the rapid transit
of a target through the Doppler beam giving a wider bandwidth of frequencies.
NOTES
• A highly focused beam (very narrow beam) will have the same effect as
a very short pulse (Figure 8.22a). Any target crossing the beam will give
a Doppler signal whose amplitude rapidly increases and then decreases
as the target moves across the beam. Any rapidly changing signal will
produce a wide spectrum in the frequency domain, as we have seen before
(Chapter 1).
• Most scanners do focus the Doppler beam at the level of the range gate.
• A very short range gate will exaggerate this effect.
The second cause of ISB results from the ultrasound beam shape as shown in
Figure 8.23. The beam is produced by a transducer with a finite aperture width. That
means that within a single beam there will be a range of Doppler angles at any one
time. A range of Doppler angles will produce a range of Doppler frequencies as given
by the Doppler equation.
NOTES
f
• The fractional spread of Doppler frequencies is given by tan D .
fD
Where Δφ is the angular spread of the aperture seen from the target.
• Larger Doppler angles of insonation (i.e. closer to θD = 90°) give greater ISB
as tan θD becomes large as θD approaches 90°
The ISB that occurs in practice will result from a combination of these two factors.
An example of ISB can be seen with a string phantom as shown in Figure 8.24. The
string moves at a single constant velocity but the Doppler signal displayed will show
a range of velocities due to ISB. ISB on velocity measurements made with a string
phantom shows that it can be as much as ±10% about the true velocity. This will affect
any peak systolic velocity measurements we make.
ϕ1 ϕ2
FIGURE 8.23 Diagram showing how intrinsic spectral broadening is caused by the finite width
of the Doppler beam which covers a range of Doppler angles.
2
(kHz)
–2
30° 40° 50° 60° 70° 80° 90°
Beam-string angle (degs)
FIGURE 8.24 The spectra obtained from a string phantom showing the spread in doppler
frequencies due to ISB at different angles of insonation. Source: Hoskins et al. [2]/with permission
of Elsevier.
NOTE
In addition to intrinsic spectral broadening the Doppler spectrum may show broadening
related to the real target having a broad range of velocities, for example, in the
presence of turbulent flow.
QUESTION: WHAT DOPPLER ANGLE SHOULD WE USE?

cfD
Recall: The Doppler equation v cos D and cos 0° = 1, cos 90° = 0
2 fT
We frequently want to measure the peak systolic velocity (PSV) of a blood flow
waveform. The Doppler angle that would give the highest fD would be that arising from
the edge of the beam, i.e. θedge as that has the smallest Doppler angle (Figure 8.25). So,
the peak velocity seen on the sonogram is that arising from the edge of the aperture θedge.
User Controls 163
θcentre
θedge
FIGURE 8.25 Definition of centre and edge Doppler angles.
However, most scanners use θcentre for the Doppler angle they use to calculate the
1
velocity. The velocity scale on the sonogram vscale is proportional to from the
cos D
Doppler equation, and the velocity is therefore overestimated with a smaller value
for cos θ at θcentre.
Therefore, in order to minimise ISB and the effects of centre beam calculation
USE AN ANGLE OF <60°
By using a standard, constant angle, variation in values obtained in measure-
ments will be minimised.
USER CONTROLS
Angle Correction
The Doppler angle cursor needs to be aligned to the vessel/flow for the velocity
scale to be correctly calibrated. Any error in positioning the angle cursor will give an
error in velocity measurements (Figure 8.26).
For example, 2° error in alignment at θD = 65° gives an 8% error in measured
velocity - therefore:
USE AN ANGLE OF <60°
Doppler Beam Steering

The Doppler beam may be steered to give a suitable angle for examining the target.
Length and Depth of Sample Volume

The target of interest is isolated by placing the sample volume (SV) over the target.
The length of the sample volume should be chosen to match the examination
required. In general it is good practice to set the SV to straddle the vessel. This will
20
18
16
14
Error in cos θ (%) 12
10
8
6
4
2
0
0 10 20 30 40 50 60 70 80
Doppler angle (θD)
FIGURE 8.26 Graph of the percentage error in cos θD at different angles of Doppler angle θD
when there is a 2° error in the measurement of θD.
minimise ISB and in the case of diseased vessels, ensure that the highest velocity is
within the SV. This highest velocity will then always be seen as the peak velocity on
the waveform.
Scale
The scale control alters the size of the Doppler waveform display within the sono-
gram image. It does so by changing the pulse repetition frequency of the Doppler
pulses used for sampling the Doppler signal. As such, adjustment of the scale can be
used to avoid aliasing of the Doppler waveform. In addition to being used to avoid
aliasing, the scale control sets the size of the waveform display. It should be adjusted
to make the waveform as large as possible without aliasing so as to ensure maximum
information is visible and the most reliable measurements can be made from placing
cursors on the waveform. (see Chapter 10).
Sweep Speed
The time scale across the width of the sonogram is set by adjusting the sweep speed.
This should be adjusted so that the information you need for the current examination
is visible and not more. For example: in most cases this might be to show 2-3 cardiac
cycles of waveform. When looking for very fast changing events, such as measuring
systolic rise time, the sweep speed should be increased. When looking for longer time
changes, such as changes in sympathetic tone, the sweep speed should be adjusted to
cover 15+ seconds (See Figure 10.13). By adjusting the sweep speed appropriately you
maximise the visibility of the information you are looking for and can make the most
reliable measurements.
Peak Velocity Envelope 165
FIGURE 8.27 Sonogram obtained from the ICA with the gain being adjusted during the time of
acquisition. This shows the change in PSV as the gain changes from too high at the left side to too
low at the right side. The centre shows the value when there was a full range of grey levels present.
Baseline Offset and Invert

When viewing bidirectional waveforms, the baseline offset should be adjusted to show
both phases of the waveform without aliasing. Inversion of the waveform may be used
to ensure forward flow is positive on the sonogram.
Doppler Transmit Power

As Doppler pulses are longer than B-mode pulses and are transmitted continually
along the same beam direction, the energy delivered to the insonated tissue is higher
than for B-mode. The transmit power should be adjusted to as low as enables the
clinical information to be obtained. Beware of over exposing sensitive tissues, e.g. the
eye and early stage foetuses. This is discussed further in Chapter 11.
Doppler Gain
The gain adjusts the overall brightness of the sonogram and should be adjusted to just
show the brightest signal as peak white. This is important as over-gain, overesti-
mates velocity and under-gain, underestimates velocities. Erroneous measure-
ments will then be made as shown in Figure 8.27.
PEAK VELOCITY ENVELOPE
Highlighting the peak velocity at each time point along the sonogram produces the
peak velocity envelope (Figure 8.28). This is a useful feature to extract for characteris-
ing different Doppler waveforms that indicate changes in physiology or pathology
in blood flow.
FIGURE 8.28 Example of a peak velocity envelope on a common carotid artery waveform using
autotrace.
Autotrace
The scanner has a facility to automatically detect the peak velocity envelope and cal-
culate measurements and indices directly from the peak velocity trace.
PROBLEM
How to detect the peak velocity automatically from a noisy real signal. Which
point defines the maximum velocity?
Figure 8.29 shows the problem of detecting the maximum Doppler frequency.
A cut-off point must be chosen. The usual way to do it is, for each instantaneous
Doppler power spectrum, the scanner calculates the velocity whereby 95% (say) of
the power (area under the curve) lies below the cut-off value and displays this point
Ideal
Real
ISB etc
E(f)
95%
100% Noise
fD
fmax
fmax?
FIGURE 8.29 The spectrum of an ideal instantaneous Doppler spectrum and a noisy real
spectrum showing the problem of determining where the peak value occurs.
on the sonogram as the peak velocity. This effectively discounts signal that is near the
noise level and therefore more likely to cause erroneous values to be calculated.
NOTES
• Autotrace will pick up noise spikes and turbulent spikes and show them as part
of the peak envelope waveform (Figure 8.30). Noise is particularly a problem
when the gain is turned up for a weak signal. In making measurements, this
can lead to erroneous values being displayed. The human eye-brain is very
good at seeing the waveform in this noise and ignoring it -USER BEWARE!
• When the peak envelope trace is displayed on the recorded waveform it
hides the true outline of the waveform making review at a later date more
difficult. It is therefore good practice to record the waveform without the
peak envelope as well, if the trace is used.
• If used, always ensure that the autotrace line does match the peak velocity
outline of the waveform and that velocity information is not being missed or
non-flow noise signal is wrongly being included.
• Using manual placement of the measurement cursor ensures you know
what is being measured.
Spike turbulence
PSV
Detected envelope
Noise
FIGURE 8.30 Showing how an autotrace waveform envelope may include spike turbulence,
when the peak systolic velocity is lower as shown (PSV), and pick up noise when there is
zero flow.
AVERAGE VELOCITY
The scanner calculates the average velocity from the ‘normalised first moment’ of
the Doppler power spectrum and may display the average velocity waveform on the
sonogram. This gives the weighted average velocity. From this, average velocities
over a period of time, e.g. 1 cardiac cycle, may then be calculated using cursors to
select the time interval for calculation.
The Normalised Fist Moment
The instantaneous Doppler spectrum is divided into a series of frequency bins 1

to n, as shown in Figure 8.31. The normalised first moment is calculated by mul-
tiplying the signal power in each bin by its bin number, then summing all these
results together and dividing that result by the sum of all the powers. A worked
example is shown Figure 8.32.
n
P n .n
first moment 1
n
P n
1
n
where x means ‘sum of x from 1 to n’
1
fD
1 n
FIGURE 8.31 Illustration of how the Doppler power spectrum is divided to allow the calculation
of the first moment to determine the average frequency in the spectrum.
0
1 2 3 4 5 6
FIGURE 8.32 Example of calculating the average Doppler frequency.
Example:
1.1 2.1 3.1 4.2 5.2 6.2 36
4 the average Dopplerr frequency fD
1 1 1 2 2 2 9
NOTES
• The average velocity will be the average from the sample volume.
• In order to get a true measure of the average velocity within a vessel, the
vessel should be uniformly insonated across its whole cross-sectional area.
This means setting a range gate that straddles the vessel. Unfortunately,
the fact that the Doppler beam is focused at the range gate means that
the beamwidth may be less than the vessel diameter and so flow near the
centre-line of the vessel is emphasised as shown in Figure 8.33. As this is
most often the fastest flow, the average velocity calculated will usually be
overestimated.
• Set the range gate to straddle the vessel to get the nearest to uniform
insonation of the vessel.
• When using calculating average velocity, different systems handle reverse
flow and absence of flow in different ways, some better than others
(Figure 8.34). For example, some ignore reverse flow, some average noise
when there is no signal – USER BEWARE!
• Beware when using average velocity that an arterial signal is not contami-
nated with a venous signal and vice versa, as both will be taken into account
when average velocity is estimated (Figure 8.34c).
Range
gate
(a) (b)
FIGURE 8.33 Placement of the range gate to straddle the vessel for average velocity
measurement (a) and the problem of focusing affecting uniform insonation within the
vessel (b).
Reverse
flow missed
(a) (b) Noise
(c)
FIGURE 8.34 Showing (a) average velocity trace on the waveform and (b) an average velocity
trace that erroneously does not include reverse flow and picks up noise when there is zero
velocity. (c) shows an extreme example of average velocity error when an arterial signal is
contaminated by a venous signal in the sample volume.
DOPPLER ARTEFACTS
As for B-mode imaging, artefacts may arise in Doppler imaging where the Doppler
waveform is affected or disturbed by factors not related to the velocities of the target
being imaged on the sonogram. These may affect the accuracy of measurements being
made and interpretation given.
Aliasing
This artefact has been covered above.
Intrinsic Spectral Broadening

This artefact has been covered above.
Doppler Artefacts 171
Wall Thump
Blood is a very poor scatterer of ultrasound (it appears black on the image) and that
is where we want to get our Doppler signal from. Arterial blood moves quickly to give
high Doppler frequencies. Immediately adjacent to the flowing blood, the vessel wall is
a very strong specular reflector that moves with the arterial pulse (Figure 8.35). It moves
relatively slowly to give low Doppler frequencies (fD < 200 Hz). The wall signal may flood
the receive amplifier to give a ‘thump’ sound on the speakers and distort the sonogram.
Strong echoes
from wall
Weak echoes
from blood
(a)
High pass
wall thump filter
(b)
(c) (d)
FIGURE 8.35 Illustration of the effect of poor penetration of ultrasound into a vessel (a) and the
effect of a wall thump filter to remove the wall signal on the sonogram(b), (c), (d).
User Control
A high-pass ‘wall thump’ filter can be applied to remove the wall signal. The level
of this filter can be selected by the user. When applied, low frequencies can be seen to
be missing from the sonogram as seen in Figure 8.35d.
Waveform Ghosting
In waveform ghosting, an inverted replica of the true waveform is seen below the zero
line (Figure 8.36). It occurs with very superficial vessels, e.g. ankle vessels, where both
forward and reverse flow are seen with in the Doppler beam aperture. Ghosting may
also occur if the gain is set very high and there is cross-talk between the forward and
reverse channels in the processing.
θR θF
Ghosting
FIGURE 8.36 Showing how waveform ghosting may occur in superficial vessels when both
forward and reverse signals are produced within the Doppler beamwidth.
References 173
Shadowing
The Doppler signal is obscured by dense tissues that cause shadowing on the image.
This is the same as for B-mode imaging.
REFERENCES
1. Guidi, G., Licciardello, C., and Falteri, S. (2000). Intrinsic spectral broadening (ISB) in ultrasound
doppler as a combination of transit time and local geometrical broadening. Ultrasound in
Medicine and Biology 26: 853–862.
2. Hoskins, P.R., Fish, P.J., Pye, S.D. et al. (1999). Finite beam-width ray model for geometric spectral
broadening. Ultrasound in Medicine and Biology 25: 391–404.
CHAPTER 9
Principles of Colour Doppler
Ultrasound
Another way to display the Doppler shift information from moving targets within the
B-mode image is to colour in the image where movement is detected. This is called
Colour Flow Mapping (CFM) or Colour Doppler Ultrasound (CDU). A map of
moving targets is produced and overlaid on top of the B-mode image [1].
Problem: We cannot use the range-gate method.
We need to interrogate each gate position for at least 4–10 ms to estimate fD
(Figure 9.1).
100 lines
Set gate
for 100
positions
FIGURE 9.1 Diagram showing how range gating could be used to interrogate an area for
Doppler information. The text describes how this would not work in practice.

175
176 Chapter 9 Principles of Colour Doppler Ultrasound
Correlation
In ultrasound signal processing there are a number of situations where it is nec-

.
essary to compare two signals to see how similar they are to one another or to
determine the difference between them. For example, CDU compares the phase
of RF signals from the field of view to determine where movement has occurred.
Speckle tracking (see Chapter 14), likewise, compares segments of the detected
signal to track movement of the speckle pattern in the image. Such changes may be
determined using correlation algorithms.
Correlation is a mathematical procedure to compare the similarity of one sig-
nal with another. When a signal is compared with a delayed version of itself it is
called autocorrelation and when it is compared with another signal it is called
cross-correlation.
The procedure can be visualised as seeing one of the signals B sliding past
the other signal A in a stepwise comparison to see where the greatest similarity is
B is slid past A in steps t −2 to t +2. The correlator output is at its maximum at
t = 0, (t0), where the two signals have the greatest similarity when compared against
one another.
In the case of detecting movement in the ultrasound image, if t0 occurs when
the time difference or position between the two signals is zero, then there has been
no movement between A and B.
3
Signal A 2 t
–1
3
2
n –3 –1
Signal B
n –2
n –1
n –0
–3 –2 –1 0 1 2 3
FIGURE 9.2 Diagram showing the principle of autocorrelation. A copy B, of signal A, is

compared with signal A in a stepwise manner to find the point of best fit.
Autocorrelation 177
If, for example, the greatest similarity t0, occurs after 5 μs or at a distance away
of 0.5 mm then, knowing the timing and location of ultrasound pulses transmitted,
we can calculate the speed or direction of movement needed to give those delayed
similarities. Likewise, the change in phase of a signal between successive pulses
can be measured.
Where the signals A and B are not exactly the same, or the signal has a lot of
random noise in it, correlation can still determine where the greatest similarity is.
In other words, it allows an estimate of the best match to be made from a noisy
signal. An example is seen in the use of coded excitation to improve signal to
noise ratio from noisy signals (see Chapter 5).
10 ms 100 100 pixels 100 seconds / image not practical!
Furthermore, we can only display one piece of information in each pixel, not a
whole spectrum.
AUTOCORRELATION
CDU uses a method of detection called Autocorrelation (Figure 9.3) [2]. This
compares a signal with a previous version of itself to detect what changes have
taken place. It produces two output values. The mean Doppler frequency fD and the
signal power.
Looking at the speckle pattern from blood, there will be a difference in phase
Δφ (≡ time delay) between the two samples due to the blood cells having moved
between the two pulses. Using the technique of autocorrelation, the Doppler
frequency shift from the moving target can be calculated from the phase shift
(see green box).
P2 Mean fD
Autocorrelation
calculator
P1(stored) Signal power
FIGURE 9.3 Showing the input and output signals from an autocorrelator.
Aside to Show that Phase Difference is Equivalent to Doppler

Frequency Shift
Consider a single target moving directly away from the probe (θD = 0).
Movement away from probe causes phase to lag (Figure 9.4).
Assume the phase difference between the two pulses is found to be 40°
Let fT 3 MHz c 1540 ms 1

0.513 mm PRF 16 kHz
Then with a difference in phase of 40° the pulse has travelled
40 / 360 0.11 of a wavelength 0.057 mm
Between the two pulses the target moves from P1 to P2 (Figure 9.5).
1 0.0567
In a time of 62.5 s the target has moved 0.0283 mm (divide by
PRF 2
2 for go and return time of the pulse between the two positions)
P1
0 40°
P2
FIGURE 9.4 Example of the phase difference between two pulses P1 and P2 due to movement
of the target between the pulses.
P1 P2
FIGURE 9.5 Showing the difference in path length for the echoes from two pulses P1 and P2
due to movement of the target between the pulses.
Autocorrelation 179
Therefore, target velocity
0.0283 1
v 453 mms
62.5 10 6
distance
using velocity
time
We now have the information about the movement of the target as a phase
difference (40°) and as the velocity in space (453 mms−1). We show the equivalence
as follows:
(a) Doppler frequency using the Doppler equation
v 2 0.453 3 106
fD 2 f 1.76 kHz
c T 1540
(b) Frequency from autocorrelator the phase changes by 0.11 period in 62.5 μs
62.5
therefore time to complete 1 cycle 567 s
0.11
1
period 567 s so fD 1.76 kHz
fD
i.e. we get the same answer from the autocorrelator as we do using the Doppler
equation to calculate the Doppler frequency from the target velocity.
What the autocorrelator is actually doing is comparing the phase from the speckle
pattern produced by the each of the ultrasound pulses regardless of whether it has
come from moving blood or stationary tissue. As the speckle pattern is likely to change
with any small movement of the probe or tissue in addition to movement due to blood
flow, the output of the autocorrelator will vary significantly from pulse to pulse. This
means the output signal is ‘noisy’ and shows random variation. The way to overcome
such a ‘noisy’ signal is to average many pulses. Each pulse pair gives an estimate of
the mean velocity and averaging improves the estimation. The number of pulses used
to form the average is called the ensemble length. Scanners use a minimum of 3
pulses but more often use ~10 pulses per colour line of data. As for pulsed Doppler,
the transmit pulses are several cycles long to give a narrow bandwidth with a well-
defined frequency (see Figure 1.18).
To form one line for the colour map, the signal is chopped into short lengths (e.g.
1.3 μs = 1 mm length in tissue) and each segment is compared with the same seg-
ment from the previous pulse (Figure 9.6). So, pulses 1 and 2 give one sample of the
Doppler frequency for each time segment, pulses 2 and 3 give a second sample etc.
Depth
Red Blue in tissue
Pixels in image colour line
FIGURE 9.6 The output from the autocorrelator for one colour line. Each consecutive pair of
returning echo pulses is broken into short segments for analysis and the output is used to colour
the displayed pixels where movement is detected.
(cf. PW Doppler sampling in Figure 8.13). By this means, over the set of 10 samples
collected for each CDU line, Doppler frequency estimates are obtained for a whole
transmit-receive line divided into CDU pixels. This is then repeated for each colour
scan line in the image.
The autocorrelator produces one value for frequency at its output for each
segment comparison. In the case of a single velocity target, this will be the Doppler
frequency of that target as shown in the box. Generally, there will be multiple targets
within the sample volume. The autocorrelator output will then be the mean Doppler
frequency of all the velocities present, i.e. the average phase shift.
COLOUR SCALE
The calculated mean Doppler frequencies are displayed using a colour scale. Change
in velocity may be shown using change in hue, saturation, or luminance. A typical
scale runs from dark red, through light red, to yellow for flow in one direction and
dark blue, through light blue, to green in the other direction (Figure 9.7). An offset
frequency is added to the signal to enable bidirectional flow to be shown, with flow
in one direction above zero and flow in the other direction below.
Yellow
Forward flow Red
Black
Reverse flow Blue
Green
FIGURE 9.7 Example of a typical colour scale used for CDU imaging.
User Controls 181
100 lines
100
Colour
pixels
(a) (b)
FIGURE 9.8 Diagram showing the dimensions for calculating the frame rate of a CDU image
(a) and the display produced from a set of colour lines (b).
FRAME RATE
The CDU frame rate depends on the number of CDU scan lines and the depth in
tissue each line interrogates, as shown in Figure 9.8.
Example: use an ensemble length of 10 pulses/line to estimate the average fD.
100 lines across the image
1
PRF for 5 cm deep scan = 16 kHz
5 13 s
16 103 prf 1
then frame rate 16 frames / second
10 pulses 100 lines
This is slow enough to look ‘jumpy’ if the probe is moved and the display may be
cosmetically improved by averaging colour frames. Frame rate can be improved by
scanning fewer colour lines and using interpolation to fill the gaps.
Because the frame rate is relatively slow it is usual to only produce a colour map
over a limited area of the B-mode image. This area is called the colour box.
USER CONTROLS
Colour Box
The size and position of the colour box may be adjusted by the user to improve frame
rate and image quality. As each line must be interrogated ~10 times to get a good
estimate of velocity, reducing the area covered by the colour box will reduce the acqui-
sition time significantly. Reducing the width of the colour box will have a greater
effect on frame rate than reducing the depth as there are then fewer sets of 10 pulses
per line to send. This is particularly important in cardiac applications where there are
fast moving targets.
Frame Averaging
FRAME AVERAGING or persistence of the colour map may be adjusted by the user.
Sampling
On some machines the number of samples (ensemble length) used to form each line
of the colour display may be chosen by the user. Figure 9.9 shows the trade off bet-
ween increasing the number of samples to give a less noisy image but with a reduced
frame rate.
(a)
(b)
FIGURE 9.9 With frame averaging turned off, (a) shows a carotid flow with an ensemble length
of 8 and a fame rate of 17 fps, (b) shows an ensemble length of 24 and a frame rate of 9 fps.
Colour Aliasing 183
θD
FIGURE 9.10 Illustration of how to ‘heel and toe’ the probe to improve the Doppler angle.
CDU AND THE DOPPLER ANGLE
As for CW or PW Doppler, the CDU signal is governed by the Doppler equation with its
cos θ term, and hence there needs to be an angle between the ultrasound beams used to
form the colour image and the blood flow being examined. This may be achieved in one
of two ways. The colour box may be steered left or right relative to the orientation of the
B-mode image. CDU ultrasound beams are then electronically steered as shown by the
box. This helps to give an acceptable Doppler angle to the blood flow seen.
The Doppler angle to the flow may be further improved by ‘heel and toeing’ the
probe as shown in Figure 9.10.
As cos 90° = 0, there is no colour signal when flow is at 90° to the CDU ultrasound
beams and blood will appear black as on the normal B-mode image.
IMPORTANT NOTE
The colour value displayed will change if either the velocity changes or the
Doppler angle changes. For example. In the case of a tortuous vessel the Doppler
angle changes along its length, so even if the flow velocity is constant along its
length, the colour shown on CDU will change.
In the illustration shown (Figure 9.11), the flow changes direction relative to the
ultrasound probe. When it is at 90° to the probe there is no flow detected (cos 90° = 0)
and the colour is black.
COLOUR ALIASING
The colour Doppler signal is built up from a series of pulse pairs that are compared
in the correlator to give an estimate Doppler frequency. As with PW Doppler, these
pulses pairs are subject to the Nyquist limit. That is, they must be less than half a
FIGURE 9.11 A CDU image of flow in a curved vessel.
FIGURE 9.12 Illustration of colour aliasing in a uniform vessel due to change of vessel angle
with respect to the ultrasound beam. High Doppler frequencies are erroneously shown at the low
end of the colour Doppler spectrum.
period of the maximum Doppler frequency apart, i.e. >2 samples per cycle, in order
to unambiguously estimate the Doppler frequency. If the time between samples is
longer than this the colour display will show aliasing. When aliasing occurs the high
Doppler frequencies will appear as low frequencies. On the colour map this appears
as a ‘wrap around’ from one end of the colour scale to the other, in a similar way to
the sonogram waveform showing ‘wrap around’ with aliasing. (see Chapter 8).
The image in Figure 9.12 shows the colour going from yellow to green in
the centre of the vessel, i.e. ‘wrap around’ due to aliasing, where there are Doppler
frequencies that are being inadequately sampled. To avoid aliasing the scale needs
to be increased, which increases the pulse repetition frequency to bring it above the
Nyquist limit.
The way to remember the colour changes seen is ‘Over the top – high velocities
and aliasing, red through black to blue – change in direction of flow’ (Figure 9.13)
User Controls 185
Yellow
Red
Change in Aliasing
Black
direction
Blue
Green
FIGURE 9.13 Showing the distinction between aliasing and a change in direction of flow.
USER CONTROLS
Scale
• The scale control alters the pulse repetition frequency of the pulses used to
form the colour lines. It may be used to avoid aliasing by increasing the PRF
above the Nyquist limit.
• It scales the detected velocities against the colour scale. In other words, it can
be adjusted so that the velocities present in the flow are mapped so as to cover
the whole colour range of the colour scale.
• Where it is required to detect fast flow, e.g. in a jet caused by disease, the scale
may be set so that for normal flow the peak colour is just below aliasing. Then,
when flow velocities exceed that level aliasing will occur and will easily be seen
on the CDU display, highlighting where fast flow is as seen in Figure 9.14.
IMPORTANT NOTE
Unlike duplex imaging, where the Doppler angle cursor is used to tell the scanner
what the Doppler angle is, in CDU the angle is unknown. Blood vessels and flow
being imaged may be at any angle to the ultrasound colour beams. The colour scale
shown on the display may have a number shown against the highest colour that
changes as the scale control is altered. This number is an indicative number only
for the scale setting. It shows the peak velocity for flow that is parallel to the colour
lines and should not be used as a velocity measure.
FIGURE 9.14 Aliasing in a straight vessel showing where a stenosis is causing increased blood
velocities.
Invert
• The colour scale may be inverted, for example, to show flow towards the probe
as red and flow away as blue, or alternatively, to show arterial flow as red and
venous flow as blue. However, in echocardiography it is conventional to
always use red towards the probe and blue away from the probe.
• The colour scale on the scanner display will always show flow toward the probe
above the zero line. This may be red or blue depending on the invert setting.
The user should always be aware which way the colour is set up, espe-
cially when looking for pathological reverse flow.
• When steering the colour box to the left and the right, the colour scale may or
may not be inverted by the scanner -USER BEWARE!
• If you are holding the probe the wrong way round in your hand the colours will
be reversed from what you expect.
When looking for reverse flow (retrograde flow) always check:
• Is the probe the right way round in my hand?

• With the box steered, is the flow toward the probe, as shown on the displayed
scale (red or blue), in the physiological direction expected?
• Only then can you confidently diagnose reverse flow.
Discrimination of Stationary Targets 187
DISCRIMINATION OF STATIONARY TARGETS
Ideally, a vessel with moving blood should have colour showing up to the wall of the
vessel, then the wall and surrounding tissue should be shown in grey scale on the
B-mode image. However, as with PW Doppler, a high pass wall filter must be used
to cut out the clutter signal from adjacent tissue that moves due to the cardiac pulse
and would otherwise give a CDU signal outside the vessel. Even with a wall filter,
discrimination of surrounding tissue is not perfect. This leads to drop-out, where
there is no colour filling in the vessel adjacent to the wall, and colour bleeding,
where colour bleeds over the edge of the vessel lumen into the wall and surrounding
tissue (Figure 9.15).
Imaging Small Vessels

The use of a wall filter also has an effect on the visualisation of small vessels. CDU is
not reliably able to visualise vessels <1 mm diameter. This is because:
• Their velocities are low and produce Doppler shifts below the wall filter cut-off
frequency.
• Small vessels themselves move at the same rate as their surrounding tissue.
FIGURE 9.15 showing how the colour filling of a vessel can bleed over the vessel wall or not fill
the lumen.
More recently viewing these very small vessels and vessels with slow moving
blood has become possible with ultrafast ultrasound and is described in Chapter 13.
USER CONTROLS
Wall Filter
As with PW Doppler, a high pass filter can be applied to the CDU signal to help
with low velocity wall movement. The cut-off frequency of this filter is adjustable
by the user.
B-Mode Priority
The colour is overlaid on the B-mode image. Related to the question of discrimination
of movement is whether the display gives preference to showing B-mode or colour.
The user can select the priority given to colour based on the level of grey displayed in
the B-mode image. This is often shown as a marker at the side of the grey scale wedge
on the display. If this is set too low on the grey scale there will be drop-out of the
colour signal as a low grey level is given priority as shown in Figure 9.16.
POWER DOPPLER (PD)
There is a second mode of displaying the colour information on the B-mode image
called power Doppler. Instead of using the directional velocity information from the
autocorrelator, the signal strength or spectral power of the Doppler signal is displayed
as a colour map regardless of frequency or direction (Figure 9.17). This is proportional
to the number of moving reflectors detected (effectively, red blood cells).
NOTES
• Because the velocity does not have to be estimated from a noisy signal, the
display of power Doppler appears smoother and is less dependent on Doppler
angle, so there is better colour filling at low velocities and close to 90°.
• It has higher sensitivity than directional CDU.
• Power Doppler is often shown using a blue–orange flame-type scale.
• Useful for viewing:
• Tortuous vessels where CDU would change colour as the vessel changes
direction. Power Doppler shows a uniform colour.
• An arterial tree, e.g. kidney or vessel malformation – vessels in many
directions show up in a uniform colour (Figure 9.18).
Power Doppler (PD) 189
(a)
(b)
FIGURE 9.16 Showing the effect of B-mode priority (a) with a low threshold for B-mode display
and (b) a high threshold. The difference is seen more clearly in the vein shown in blue.
PD
f0 fD
FIGURE 9.17 Power Doppler shows the strength of the Doppler signal which is equivalent to the
area under the curve of the Doppler spectrum.
FIGURE 9.18 Power Doppler image of a foetal circle of Willis. Note that vessels with flow in
different directions are shown uniformly. Source: Courtesy: Esaote.
NOTE
Some machines improve the quality of the directional colour display by using the
power Doppler information and colouring it in with the directional information.
A variation on the conventional way to display CDU and PD images is to use the
velocity (CDU) or power (PD) scale to give a ‘3D’ appearance to the displayed image
as shown in Figure 9.19. The higher values of the colour velocity are used to brighten
the displayed image. By enhancing the display of flow in this way the vessel gains a 3D
effect as the fastest flow tends to be along the centre line of the vessel with slow flow
adjacent to the vessel wall. This is seen in Figure 9.19b.
CDU ARTEFACTS
As for B-mode artefacts, some CDU artefacts can be useful.
Aliasing
This can be a useful artefact as it can show where velocity increases at a stenosis in a
vessel -hence it can show where the PW Doppler sample volume needs to be placed
to measure the highest velocity. See Figure 9.14.
Shadowing
CDU is subject to shadowing from dense/reflecting structures in tissue causing colour
dropout (Figure 9.20)
CDU Artefacts 191
(a)
(b)
FIGURE 9.19 ‘Radiant flow™’ enhancement of CDU display. (a) shows how the colour scale is
used to give a ‘3D’ effect to the displayed CDU map. (b) shows an example of the umbilical vessels.
Source: GE.
FIGURE 9.20 Carotid artery showing dropout in the colour distal to a shadow casting plaque.
Wall Discrimination
NOTE
Because of dropout and colour bleeding near vessel walls, colour should not be
used to make diameter measurements of vessels. Instead, use a good quality
B-mode image where the walls are clearly imaged.
Flash Artefact
This is caused by probe movement on the skin causing colour signal to be generated
from the tissue moving relative to the probe. It appears as colour flashes across the
whole image. Algorithms to suppress flash artefact mean that it is much improved in
modern scanners.
Mirror Image Artefact

As for B-mode, colour can be shown where the beam has reflected from a surface in
the body, e.g. from bone at the ankle for the tibial vessels, from the lung for subclavian
artery (Figure 9.21).
Twinkle Artefact
Calcified vessels and stones can show a colour ‘twinkle’ artefact that is not related to
blood movement (Figure 9.22). Its origin is not clearly understood but may be similar
to comet tail artefact in B-mode, i.e. reverberations generating a spurious Doppler
signal. It appears as colour sparkling at points behind the calcification. Its occurrence
is sensitive to machine settings [3].
Temporal Artefact
The frame rate for CDU is relatively low and at 20 fps it takes 50 ms to obtain 1 frame.
During the systolic rise time arterial flow velocity can increase by 1 ms−1 in 90 ms.
In other words, in the time it takes to obtain one frame, the flow can significantly
change. This gives rise to a temporal artefact whereby one side of the colour box can
show a different colour than the other, even though it is one uniform straight vessel
being viewed (Figure 9.23).
COLOUR SENSITIVITY
It can be difficult to get colour filling in a deep vessel when there is poor visualisa-
tion or flows are very low. CDU needs a Doppler angle. This can be achieved either
by beam steering or by ‘heel and toeing’ the probe. Recalling from Chapter 3 that
SCA
Lung
Reflected
(a) SCA
(b)
FIGURE 9.21 Reflection of the subclavian artery (SCA) seen within the lung after reflection
from the top surface of lung.
FIGURE 9.22 An example of twinkle artefact from a stone in water showing that its sonogram is
just noise. Source: Hirsch et al. [4]/Permanyer.
FIGURE 9.23 Temporal artefact showing a change in velocity along a common carotid artery due
to the time taken to acquire the image. Frame rate was reduced to 3 fps to clearly show the effect.
the transducer has its greatest sensitivity when the beam is straight ahead with no
steering, sensitivity is improved by steering straight ahead and then using ‘heel and
toe’ to give a Doppler angle to the vessel.
For low flows, turn the colour scale down and turn the gain up until colour noise
is just apparent to achieve greatest sensitivity.
PRESETS
Manufacturer presets for different examinations will alter the CDU settings discussed.
As the user, it is important to make sure the settings are appropriate for the patient
in front of you and that you re-adjust the settings as necessary to achieve optimal
CDU imaging.
COLOUR M-MODE
M-mode enables tissue movement along a single line of sight to be appreciated over
a period of time (Chapter 5). Using colour Doppler on an M-mode display enables
changes in blood velocity and direction along that line to be visualised. As only one
scan line is being imaged there is high temporal resolution and the technique is useful
within echocardiography (Figure 9.24).
TISSUE DOPPLER IMAGING (TDI)
Tissue Doppler imaging is used in cardiac examinations. It is based on the prin-

ciple that blood moves quickly whilst the myocardium moves relatively slowly.
Tissue Doppler Imaging (TDI) 195
FIGURE 9.24 A colour M-mode display of flow across an aortic valve with regurgitation.
Source: Courtesy: Chris Eggett.
Low pass filter
Tissue doppler signal

PD
Blood doppler signal
10 100 1000 10000 fD
FIGURE 9.25 Spectrum showing the low pass filter used to isolate the tissue Doppler signal.
Instead of using a high-pass filter to filter out the wall movement, the reverse is done
(Figure 9.25). A low-pass filter is used to filter out the blood signal and allow the
cardiac wall movement to be shown in colour (Figure 9.26). This enables areas of
myocardial ischaemia to be identified as the affected walls move paradoxically to
normal myocardium.
TDI will detect any tissue motion, both passive, where there is inactive myocar-
dium or due to tethering of tissue to a moving region, and active motion due to muscle
contraction.
If all points in a region of the image move together, we have displacement of the
target. If different points move at different velocities, we have distortion or defor-
mation of the target, and the target is changing shape (Figure 9.27).
(a)
(b)
FIGURE 9.26 A tissue Doppler image of a 4 chamber view showing movement in the normal
heart in (a) diastole and (b) systole. Source: Courtesy of Chris Eggett.
t t
Δd Δd
t0 t0
P1 P2 P3 P4 P5 P1 P2 P3 P4 P5
(a) (b)
FIGURE 9.27 Showing position of points in a structure P1 to P5 at times t0 and t. (a) All points
have moved the same distance Δd and we have displacement of the structure. (b) The points
have moved with different velocities and their positions at t are at varying distances and we have
distortion or deformation of the structure.
Myocardial Strain Imaging 197
MYOCARDIAL STRAIN IMAGING
A further development from simple observation of cardiac motion with TDI is to mea-
sure myocardial strain and strain rate. This can give detailed information on how the
myocardium is performing and differentiate active versus passive movement [5].
Strain
Strain ε is the increase or decrease in length of a dimension of the myocardium over
a reference length L0, for example, the length at end diastole in the cardiac cycle, in
response to an applied force (Figure 9.28). This is also called the Lagrangian strain.
Lt L0
100
L0
It is a dimensionless number expressed as a percentage, positive for an increase in

length and negative for a decrease in length, where Lt is the length at time t.
Strain Rate (SR)

Strain Rate (SR) is the rate of change of strain, that is how quickly the change in
length is taking place.
L 1 v 1
SR . equals the time gradient of unit : s
t L0 t L0
SR is positive on lengthening and negative on thinning or shortening. It is a mea-

sure of the velocity of wall movement from point to point.
Three dimensions of strain are considered as shown in Figure 9.29:
1. Longitudinal strain L (base – apex of ventricle)

2. Radial strain R (thickening of myocardium)
3. Circumferential strain C (change in length as the circumferential line con-
tracts and relaxes)
ΔL
0 L0 Lt
FIGURE 9.28 The dimensions for measuring strain ε.

FIGURE 9.29 The three dimensions of strain within a cardiac chamber considered in cardiac
examinations.
VA
A•
B•
VB
FIGURE 9.30 Points A and B selected to measure the velocity difference using TDI for
calculating SR.
Using TDI to Measure Strain

Two regions of interest (ROI), A and B a distance d apart, are selected on the image
(Figure 9.30). Using TDI, the Doppler velocity is measured at points A and B to obtain
the velocities vA and vB at these points. The difference in velocities is the velocity gra-
dient Δv from which the SR can be calculated:
v vB vA
SR
d d
Speckle Tracking Echocardiography (STE) 199
SR
ε
t
t0 t
FIGURE 9.31 Strain ε as the summation of change in strain rate over time.
The strain may then be calculated from the summation of the strain rate over time:
t
SR.dt
t0
i.e. the area under the SR curve over time as shown in Figure 9.31.
Using TDI to calculate strain and strain rate has a number of limitations.
• It is dependent on the Doppler angle of incidence between the ultrasound

beam and the myocardial motion. If this is >20° the strain and SR will be sig-
nificantly underestimated.
• It requires an acquisition rate of >130 frames per second.
• It needs a good B-mode image with optimal wall-TDI alignment.
• The Doppler scale must be adjusted to avoid aliasing.
• There is a trade-off between the size of the ROI for A and B which determines
the SNR and spatial resolution.
TDI derived strain and SR has a typical variability of 10–15%.

These limitations of TDI measured strain and SR are largely overcome by the use
of speckle tracking from the B-mode image to measure strain.
SPECKLE TRACKING ECHOCARDIOGRAPHY (STE)
A series of 2-D B-mode images from across the cardiac cycle are stored (typically, 3
cycles are obtained) and the movement of the speckle pattern is tracked across the
images using a technique such as cross-correlation between the images (Figure 9.32).
Movement may be detected in both lateral and axial directions to enable speckle paths
to be identified and velocity vectors for actual direction of movement to be calcu-
lated [6] (see Chapters 13 and 14 for descriptions of vectors and speckle tracking)
From the velocity vectors, strain and SR may be calculated along the vector as for
TDI velocities described above. Analysis is performed on a segment by segment basis
as well as global longitudinal values.
FIGURE 9.32 Illustrating the principle of speckle tracking to locate the change in position of
speckle ‘markers’ in the B-mode image.
Torsional motion (twisting) may be calculated by measuring rotations (circum-

ferential motion) at two levels. Torsion is then the net difference between the two
rotations in degrees/second.
STE has a number of advantages over TDI strain imaging.
Pros:
• Velocities are not dependent on Doppler angle.
• Deformation may be measured in more than one direction.
• Higher spatial and temporal resolution than TDI.
• Optimal frame rate for B-mode acquisition is 50–70 fps.
• Better reproducibility than TDI method.
• Fewer artefacts seen than with TDI.
• Less sensitive to signal noise.
Cons:
• Needs high quality B-mode images which may limit use in some patients.
• Signal may be lost with translational motion of the heart out of the scan plane.
This may be avoided where 3-D tracking data is available.
Speckle Tracking Echocardiography (STE) 201
• Processing is off-line and requires manual input to identify and outline the
endocardial border on one image in order for analysis of the whole image set to
be performed. Newer techniques using advanced algorithms are able to over-
come these limitations as discussed in Chapter 12.
• Because different algorithms are used by different vendors, measurements
should not be compared across different manufacturers scanners.
STE requires good quality B-mode images with optimal setting of gain, focus and
image width and depth centred on the cardiac chamber of interest. Breath holding
by the patient will minimise extraneous movement which can create artefacts. In
particular, the endocardial and epicardial borders should be clearly delineated. A
good quality ECG signal is also required for gating.
STE Display
The velocity, strain and strain rate data may be displayed in a number of ways. They
may be shown graphically as a change over the cardiac cycle, as a colour mapping
over the B-mode image or as a segmental display known as a Bull’s Eye Plot that dis-
plays peak systolic strain in each segment of the myocardium (Figures 9.33 and 9.34).
(a) (c)
(b) (d)
FIGURE 9.33 Regional strain around the normal left ventricle. Panel (a) shows the points from
which the strain has been calculated. Panel (b) shows the regional strains at peak systole with each
region colour coded. Panel (c) shows the strain for each region over the cardiac cycle, and panel (d)
shows a heat map of the strain over time for the measured points going round each region versus
time on the x-axis. Source: Courtesy of Chris Eggett.
(a)
(b) (c) (d)
FIGURE 9.34 Segmental strain traces recorded from apical 3-, 4-, and 2-chamber views,
respectively (a); bull’s eye plots displaying midwall peak systolic strain (b); time to peak longitudinal
strain (c); and post-systolic index (percentage of post-systolic increment over systolic strain) (d) in a
patient with a left circumflex infarct. Source: Collier et al. [8]/with permission of Elsevier.
REFERENCES
1. Evans, D.H., Jensen, J.A., and Nielsen, M.B. (2011). Ultrasonic colour Doppler imaging. Interface
Focus 1: 490–502. https://doi.org/10.1098/rsfs.2011.0017.
2. Hedrick, W.R. and Hykes, D.L. (1995). Autocorrelation detection in color Doppler imaging: a
review. Journal of Diagnostic Medical Sonography 11: 16–22.
3. Kamaya, A., Tuthill, T., and Rubin, J.M. (2003). Twinkling artifact on color Doppler sonography:
dependence on machine parameters and underlying cause. American Journal of Roentgenology
180: 215–222.
4. Hirsch, M., Palavecino, T., and León, B. (2011). Color Doppler twinkling artifact: a misunder-
stood yet useful sign. Revista Chilena de Radiología 17: 82–84.
5. Nesbitt, G.C., Mankad, S., and Oh, J.K. (2009). Strain imaging in echocardiography: methods and
clinical applications. The International Journal of Cardiovascular Imaging 25: 9–22. https://doi.
org/10.1007/s10554-008-9414-1.
6. Johnson, C., Kuyt, K., Oxborough, D. et al. (2019). Practical tips and tricks in measuring strain,
strain rate and twist for the left and right ventricles. Echo Research and Practice 6: R87–R98.
https://doi.org/10.1530/ERP-19-0020.
7. Blessberger, H. and Binder, T. (2010). Non-invasive imaging: two dimensional speckle tracking
echocardiography: basic principles. Heart 96: 716–722.
8. Collier, P., Phelan, D., and Klein, A. (2017). A test in context: myocardial strain measured by
speckle-tracking echocardiography. Journal of the American College of Cardiology 69: 1043–1056.
https://doi.org/10.1016/j.jacc.2016.12.012.
CHAPTER 10
Making Measurements
A great deal of the usefulness of ultrasound imaging arises from measurements that
can be made from the images. Values obtained may be compared to classify normal
from abnormal and to monitor changes over time. It is important, therefore, to under-
stand the limitations on measurements made and to consider the errors that may
arise. There are a number of key concepts relating to measurement which we consider
followed by a consideration of how they apply to making measurements on ultra-
sound images and Doppler waveforms. We begin with a discussion on accuracy and
precision (Figure 10.1).
Measured
value
True
value
Accuracy
Precision
FIGURE 10.1 Graphical display of a range of measurements showing the accuracy and precision
of a measured quantity.

203
204 Chapter 10 Making Measurements
ACCURACY
The accuracy of a measurement is its closeness to the true value.
EXAMPLE 1
If you ask a friend the time and they look at their watch and say my watch says
10.00am and then immediately you see a radio clock tied to atomic time and see
that it is in fact 10.05am, the time they gave you is inaccurate by 5 minutes. That
may or may not matter to you, depending on what you need to know the time for.
The time they gave is right to within 5 minutes. The radio clock, on the other hand,
is always telling the true time as it is tied to an atomic clock from which the passage
of time is defined.
EXAMPLE 2
If you had a homemade ruler where you had drawn the marks on the ruler by hand
and made each centimetre gap approximately 1 cm but on average you had made
them 2 mm too large, then any measurements you made with the ruler would be
inaccurate and the error would be larger for larger distances. The error for any one
measurement would vary in a random way but the average error would be 20%
too large.
PRECISION
Precision is the range of variation of measured values around the average mea-
sured value.
EXAMPLE 3
If you asked your friend the time at several points during the week and every time
the answer they gave you was exactly 5 minutes behind the radio clock, then their
watch could be considered precise, though inaccurate. Although, once you knew
there was a 5-minute offset, you could add 5 minutes and obtain the accurate time.
This may be compared with the situation where it was 5 minutes behind, then
3 minutes behind then 6 minutes behind etc. but on average was 5 minutes slow.
In this case the precision is ±2 minutes (say).
Another way of looking at precision is to ask how closely can I specify the
value I measure to the actual value I believe I have measured – which I assume to
be the true value.
Reproducibility 205
EXAMPLE 4
If you have a ruler that you believe to be accurate – the distance marks are spot
on – but which is only marked in centimetres, then you can estimate the gap bet-
ween the marks to maybe a quarter of a centimetre. However, you would not be
justified in quoting a value to 3 decimal places as you cannot be that precise in your
measurement. The precision of a measurement will limit how finely the value can
be meaningfully specified.
Accuracy and precision are closely connected but are different concepts. The
atomic clock is both accurate and precise (accurate to 1 second in millions of years
and precise to 1 part in 1010).
Example 1 is precise but inaccurate, Example 2 is inaccurate and imprecise and

Example 4 is accurate but imprecise.
These examples illustrate that precision may be limited by the finesse
of my measuring device or be limited by random events occurring in the
measuring process.
HOW ACCURATE OR PRECISE DO WE NEED TO BE?
Depends on:
• What you want to use the data for, for example, comparing kidney size or dating
a pregnancy require different levels of precision.
• What you are comparing it with, e.g. CT, MRI, measuring tumour growth from
a previous examination, standard normal values.
• What the normal range of variation is – a large normal range does not require
a highly accurate measurement to confirm normality.
Accuracy and precision become crucial when you are pushing the limits of what
ultrasound imaging is capable of resolving or when small errors make a significant
difference in diagnosis and clinical decision making. For example, measuring nuchal
translucency, or intimal thickness in arteries.
REPRODUCIBILITY
For the purposes of making measurements in a clinical setting, a closely related con-
cept is that of reproducibility. If I make a measurement, do I get the same value when
I repeat that measurement – intraobserver variability – or when someone else
makes the same measurement– interobserver variability?
Reproducibility is likely to be poorer when the same target is measured on dif-

ferent scanners and caution needs to be exercised when comparing measurements
made across scanners. Such measurements require the scanners to be calibrated
against standard test objects to ensure accuracy and precision are within accept-
able limits.
SYSTEMATIC AND RANDOM ERRORS
Systematic Errors
Systematic errors produce a consistent difference from the true value. They can
be corrected if the cause and value of the error are known. For example, times
obtained from the watch that is always 5 minutes wrong can be corrected to give
the true time.
Random Errors
Random errors give a lack of precision to any measurements we make. They may
be minimised with careful attention to technique but will always be present to some
degree. Random errors cannot be corrected but can be reduced by making multiple
measurements and taking the average value.
Random error reduces by n of the number of measurements n made. So, for
example, the average of 4 measurements will halve the random error.
Random errors will result from:
• Technique used to make the measurement.

• Intrinsic limits of ultrasound scanning, e.g. image resolution, geometric
artefacts.
• Signal noise.
• Target movement.
• Introducing variables such as comparing measurements across machines.
ULTRASOUND MEASUREMENTS IN PRACTICE
We consider making measurements from images first then look at Doppler

measurements.
The accuracy of measurements made from ultrasound images depends on the
physical limitations of the ultrasound scanner and the process of image creation,
and on the sonographer making the measurements. These two may be considered
separately but the overall result will depend on the combined errors and limita-
tions of both.
Physical Constraints 207
PHYSICAL CONSTRAINTS
The accuracy of measurement will depend on:
• Accurate scaling of the image

• Caliper accuracy
• Digital quantisation
• Image resolution
Scaling
In the lateral direction the scale relates to the physical width of the probe. In the case
of a linear array, with no beam steering at the edge of the field of view, this is simply
seen as the width of the active transducer array (Figure 10.2).
In the axial direction the scale depends on the system velocity. The positioning
of an echo in the axial direction depends on the go-and-return time of the pulse which
is determined by the speed of sound in tissue. For soft tissue the scanner assumes a
system velocity of 1540 ms−1. In reality the speed of sound varies by ±5% around this
value. This variation will lead to misplacement of echoes and limits the accuracy of
any axial measurement. In some clinical situations a more precise system velocity
is used that better matches the tissue being measured. For example, in ophthalmic
scanning the speed of sound for the cornea and lens (1641 ms−1) and aqueous and
vitreous humour (1532 ms−1) may be used instead of 1540 ms−1. More recent develop-
ments that enable speed of sound correction improve the accuracy of measurement
(see Chapter 13).
Caliper Accuracy
Caliper accuracy is the accuracy of the indicated caliper measurement compared
with a known measurement in a 1540 ms−1 tissue mimicking phantom. It is generally
<1–2% (1 mm in 10 cm). Endoscopic probes tend to be slightly less accurate.
cm
2d
c
FIGURE 10.2 The dimensions of an image produced by a linear array.

Δ0.2 mm
FIGURE 10.3 The quantisation error of a digitised caliper measurement.
Caliper precision is limited by quantisation error (Figure 10.3). As a caliper

is moved from one pixel to the next there will be a discrete change in the value of the
reading given. For example, 4.6–4.8 mm with Δ = 0.2 mm. The exact value of this will
depend on the pixel density, image depth and zoom used. It is therefore not mean-
ingful to quote point to point linear measurements to a greater precision than this
level of precision.
Image Resolution
The lateral and axial resolution of the image will limit the detail seen in the image
and therefore limit the precision of measurements that may be made. Typical values
for resolution are:
Axial resolution ~1.5 mm at 3 Mhz

~0.6 mm at 7 MHz
Lateral resolution depends on beamwidth – typically 1.5–3 mm

The definition of edges in the image will also be affected by artefacts such as:
• Reverberation between closely separated boundary layers.

• Poor visualisation due to curved surfaces running in an axial direction.
• Clutter from slice thickness artefact, grating lobes and multiple reflections.
• Poor visualisation and image distortion due to patient habitus and overlying
fatty tissue causing refraction and beam defocusing.
These factors are all intrinsic to the image creation and display process within
the scanner.
SONOGRAPHER-BASED CONSTRAINTS
When making measurements the accuracy and precision of measurement will

also be affected by sonographer choices when operating the scanner and placing
the calipers.
Primary choices:
• Probe frequency – higher frequency gives finer resolution.

• Focus – set the focus to the level of the target of interest for best resolution.
• Compound imaging – improves the definition of curved surfaces.
• Zoom – write zoom before freezing the image gives more information, zoom-
ing the frozen image simply enlarges the information already present.
PRINCIPLES FOR MAKING RELIABLE MEASUREMENTS
Target Visualisation
Ensure the target is as clearly visualised as possible. Boundaries appear sharpest when
viewed perpendicular to the ultrasound beam axis (Figure 2.3). Caliper placement
will always be more reliable if placed on an interface imaged perpendicular to the
beam and the ideal angle of approach to the target should be chosen so that this is the
case. Compound imaging will improve the visualisation of curved surfaces.
Frame Freeze
The ultrasound probe must be held completely still when freezing the image to avoid
blurr in the image (see Figure 7.27). This is especially the case when a lot of frame
averaging (persistence) or multiple transmit focal zones (reducing frame rate)
are being used. It may necessitate asking the patient to breath hold whilst freezing
the image.
Use First Interface

It is generally a good principle to use the ‘first interface’ for caliper placement in mak-
ing measurements (Figure 10.4). Immediately distal to the first interface the image
may be confused with reverberations arising from first interface layers and slice
thickness artefacts.
Measurements Protocols
In order to obtain reliable and robust measurements for clinical decision making
it is good practice to use measurement protocols to ensure everyone making the
(a) (b)
FIGURE 10.4 (a) Diagram of caliper placed on first interface of the target to be measured.
(b) Enlargement of liver-kidney boundary on an ultrasound image.
FIGURE 10.5 Radiograph of an aortic aneurysm showing examples of possible diameter

measurements.
measurement is making it repeatably in the same way. This is particularly important

where fine or difficult measurements are being made. Examples of protocol driven
measurement where image plane and caliper placement are specified include foetal
biometry, echocardiographic measurements and aortic aneurysm screening. [1–3]
As an example of the problem of deciding what to measure, the radiograph in
Figure 10.5 illustrates a number of ways in which calipers could be placed to gauge
the size of an aneurysm. A protocol should specify which measurement to use for
consistency of reporting.
Actual Measurements
As shown in Figure 10.6, the measurement of distance between the two boundaries
may vary between AF and CD depending on where the you place the calipers. Dif-
ferent sonographers are likely to place the calipers in slightly different positions,
even given the principle of first interface placement. This is especially the case when
the boundary is less distinct. This variation will give rise to a random error in the
measurement.
In addition to the random variation in making such measurements, you will
have a personal systematic error in the measurements you make. That is, you will
generally tend to place the calipers in a particular spot within the image of a com-
plex boundary. That will be slightly different to someone else, in other words, it is a
personal preference particular to you.
Factoring Variability
These scanner-based factors and personal factors all contribute to the random error
associated with any particular measurement. They give rise to intra-and inter-operator
variability.
Random errors reduce as n of the number of measurements made, i.e.
averaging 4 independent measurements of the same thing reduces the
error associated with the measurement by half. Independent measurement in
this case would be unfreezing the image, acquiring it again and re-making the
measurement.
(a) (b)
FIGURE 10.6 Diagram showing the possible range of caliper measurements across two
interfaces (a). Illustration of such a pair of interfaces in a blood vessel (b).
For an area or volume measurement V, the random errors v, combine as shown:
2 2 2
v a b c
V A B C
where a/A, b/B, c/C are the error and dimensions contributing to the volume V.
Ultrafine Measurements
Measurements of fine structures at the limit of resolution in the B-mode image may
be successfully made where averaging of the measurement can be used. This may
be by making multiple measurements and averaging. Or it may be by using some
other averaging technique. An example of this is in the measurement of intima-media
thickness (IMT) and flow mediated vessel dilatation, where the changes observed are
of the same order as the axial resolution of the scanner.
For reliability and repeatability in all such ultrafine measurement, a scan
protocol must be followed that includes a comprehensive set of machine settings to
be used such as gain, zoom, frame averaging, position of focus, greyscale mapping etc.
In the case of IMT a longitudinal image of an artery is obtained such that the vessel
wall with an intima-media boundary is visualised across the image (Figure 10.7). A
dedicated linear cursor is placed along the intima-blood boundary and another along
the intima-media boundary. Even though this boundary may not be sharply distinct
along its length, the line of it can be seen and the linear cursor effectively averages the
line of the boundary across the image. Some scanners provide an automated detection
of the boundaries for IMT measurement.
10 mm
(a) (b)
FIGURE 10.7 Diagram showing positioning of cursors for IMT measurement (a) and (b) an image
with automated detection of the boundaries. Note the scale. The measurement is at the limit of
ultrasound resolution. Source: Kastelein and de Groot [4]/with permission of Oxford University Press.
Ultrafine measurements of changes in vessel diameter with flow mediated

ilatation or the cardiac cycle may also be made automatically using speckle tracking
d
techniques. The RF echoes from the proximal and distal vessel wall-blood boundaries
are tracked using autocorrelation to determine movement (see Chapter 9).
NOTES
• An ECG signal may be used to gate the measurement to the cardiac cycle.
• The measurement should be repeated several times and results averaged to
improve precision.
MEASUREMENT OF CIRCUMFERENCE, AREA, AND VOLUME
Measurements involving more than one linear dimension are frequently required in
ultrasound examinations. A number of approaches may be used in making such mea-
surements. Each involve assumptions that result in the measurement being more or
less accurate.
Direct Measurement
Area and circumference may be measured by directly tracing out the boundary to
be measured. Using a tracker ball or other interface, the caliper is traced around the
boundary, and the delineated circumference or area enclosed is calculated.
• Relies on hand-eye coordination of the sonographer tracing round the area.

• For circumference errors of overshooting or undershooting the boundary add
to the length increasing overall error.
• For area measurement, overshooting and undershooting tend to even out.
Assume a Simple Shape

It is often sufficient to assume a simple shape such as an ellipse and match that shape
visually to the target to be measured.
• Many targets approximate an ellipse very well.

• Reduces input required by the sonographer.
• Good reproducibility.
In the case of an elliptical shaped target the estimate may be made by measuring
the long and short axis of the ellipse so the scanner can calculate the values, e.g. for
foetal head measurements (Figure 10.8). The placement of an elliptical cursor on the
target will perform this calculation.
B
A
FIGURE 10.8 The dimensions used when assuming an ellipse for calculating area or
circumference.
L
H
FIGURE 10.9 The measurements to be made for calculating bladder volume.
Use a Simplified Model of the Target

By using a simplified model of a complex shape, an estimation of the circumference,
area or volume may be made using just a few linear measurements.
An example of such a model may be used to measure bladder volume
(Figure 10.9) [5].
One model used is that of a prolate ellipsoid. The formula used is
V 0.52 length width height
and volume = V ±25%.

This degree of imprecision may be considered adequate for this measurement.
Slicing Models
A closer estimate of the true volume may be made by measuring a series of slices
through a volume and calculating the volume as a series of trapezoid volumes added
together as shown in Figures 10.10 and 10.11.
Surface Segmentation and Automated Methods

With the advent of 3D ultrasound and artificial intelligence algorithms, it is possible
to automatically detect surfaces within the image data and segment or define the
volume to be measured (see Chapter 12).
A1 A2
(A1 + A2)
V= ×h
2
FIGURE 10.10 The method of calculating volume using trapezoid slices.
(a) (b)
(c)
FIGURE 10.11 Illustrating the principle of deriving the bladder volume from a series of image
slices. (a) A B-Mode image slice, (b) illustrates the 3D position of the slices to be used and (c)
shows the reconstructed bladder volume. Source: Images courtesy of Dr Graham Treece.
DOPPLER WAVEFORM MEASUREMENTS
Doppler Set-Up
When observing Doppler waveforms and in order to make reliable velocity measure-
ments that maximise the available information in the sonogram, it is important to set
up the Doppler display correctly.
Image/Sonogram Size
When viewing Doppler waveforms, the sonogram is the main focus of interest and
the sonogram display should be the major part of the screen display with the B-mode
image reduced in size to be used for guidance only.
Scale
• Set the velocity scale so that the waveform fills the space. You then have the
largest size of display from which subtle changes in waveform may be seen and
more precise measurements made (Figure 10.12).
• Set the time-base 4–6 seconds or to see what you want to see and no more. For
example, to measure systolic rise-time (SRT) a faster time base is needed (2–3
seconds) and to look at autonomic changes in diastolic flow a longer time base
is needed (e.g. 15 seconds) (Figure 10.13)
NOT
FIGURE 10.12 The correct scaling to use when viewing a Doppler waveform.
FIGURE 10.13 Using a 15 second time base to illustrate the cycling of sympathetic tone in
an artery.
Doppler Waveform Measurements 217
Gain
Doppler gain should be adjusted so that the greatest intensity seen in the waveform
just reaches peak white (see Figure 8.26)
NOTE
Over-gain, overestimates velocity
Under-gain, underestimates velocities
Freeze the Image

Whilst acquiring Doppler waveforms, the B-mode image should be frozen to avoid
time sharing Doppler with B-mode. This gives much better-quality waveforms as it
uses longer, more frequent samples, giving better frequency and time resolution on
the sonogram.
Weak Signals
To improve imaging where the signal is weak:
• Reduce beam steer toward ‘straight ahead’ and ‘heel and toe’ the probe to
obtain a good Doppler angle (recall Chapter 3 – a forward looking beam is more
sensitive) See Figure 9.9.
• At very shallow angles there is poorer penetration of ultrasound into the vessel
lumen and errors in velocity measurement can occur. Therefore, use an angle
of >45° (Figure 10.14)
Doppler Angle
The greatest source of error in making measurements from Doppler waveforms is a
misalignment of the Doppler angle cursor. At an angle of 60° a 2° error in cursor
alignment will produce an error in velocity measurement of ~6% (see Figure 8.25).
The angle cursor moves in discrete steps, e.g. 2° steps, giving a quantisation error
(Figure 10.15). Accurate alignment of the angle cursor is therefore vital for reliable
velocity measurements.
FIGURE 10.14 Reflection of the Doppler beam at the vessel wall with a shallow angle of
insonation giving poor penetration into the vessel.
2° steps
FIGURE 10.15 Typical quantisation error in positioning the angle cursor.
FIGURE 10.16 Doppler waveform from the common carotid artery with the artery correctly
aligned along its length to show the vessel walls and intima and the Doppler angle set correctly.
For cardiac work
• It is usually possible to obtain an alignment for the Doppler signal that is in

line with the flow being measured, e.g. through a valve. In this case the Doppler
angle is 0°.
• When looking at flow jets the Doppler beam must be accurately aligned with
the jet flow for a 0° angle assumption to be valid (see Figure 12.15).
For vascular work, in order to make accurate measurements in arteries and

veins the following should be observed:
• Align the vessel so as to show the longest length of sharply imaged vessel
walls across the width of the B-mode image, as shown in Figure 10.16. Then
you know you are lined up with the vessel axis and in line with the flow in
most cases.
• When the vessel is very tortuous then use the tangent to the curve of the vessel
at the point of measurement to position the angle cursor (A) (Figure 10.17).
Waveform Indices 219
FIGURE 10.17 Diagram showing the correct alignment of the angle cursor in the case of a
curved vessel (A) and an eccentric jet (B).
• Where disease causes an eccentric jet of flow then align the angle cursor to the
line of the jet (B).
An ideal Doppler angle is 45–60° – (or fix at 60° to minimise variability due to
angle effects)
The Doppler sample volume should be wide enough to ensure that the peak
velocity in the flow is captured. To ensure this and reduce intrinsic spectral broad-
ening, a sample volume that straddles the vessel is recommended.
Having obtained the correctly aligned best waveform possible, measurements
may be made using the manual cursor or auto trace as discussed in Chapter 8.
WAVEFORM INDICES
These were used extensively in the early days of CW scanning to avoid the ‘unknown
angle’ problem (i.e. no B-mode image, just the Doppler sonogram). They are still used
in some situations. [6]
Pulsatility Index (PI)

Pulsatility Index (PI) – gives a measure of how pulsatile a waveform is.
peak peak
PI
average of peak envelope
NOTE
If the equation is written out in terms of the Doppler equation to calculate the
velocities, then cos θ appears on top and bottom so cancels out, so PI is independent
of Doppler angle.
As the waveform becomes less pulsatile the PI reduces in value. (PI value range ≥0)
Examples as shown in Figure 10.18:
80
Average velocity of
peak envelope
P-P
8
0
–20
(a)
40
P-P
20
10
0
(b)
FIGURE 10.18 Examples of pulsatility index parameters. (a) 80/8 = 12.5, in a normal femoral
waveform and (b) 40/20 = 1.5, in a damped waveform.
Resistance Index (RI)

Resistance index (RI) – also known as Pourcelot’s RI – gives a measure of
peripheral resistance
A B
RI
A
Diastolic flow is a good measure of peripheral resistance and as resistance reduces

B increases relative to A and the value of RI decreases. (RI value range 0–1)
NOTE
Again, cos θ appears on top and bottom so cancels out, so RI is independent of
Doppler angle.
Examples as shown in Figure 10.19:
A A
60 60
B
45
B
10
(a) (b)
FIGURE 10.19 Examples of Resistance index parameters. (a) (60 − 10)/60 = 0.83, in a normal
common carotid waveform and (b) (60 − 45)/60 = 0.25, in a low resistance fistula waveform.
A/B Ratio
A/B ratio – is similar to RI with the same A and B values uses as shown above. This
index is still used in obstetrics
COLOUR DOPPLER ULTRASOUND
CDU is used to show where there is flow and can give a good indication of where to
place the Doppler sample volume to obtain the Doppler waveforms. However, it is of
limited use in making measurements in its own right. Velocity measurements cannot
be made from CDU as the flow angle is not known.
CDU should only be used to make measurement of patent lumen diameter when
there is an inadequate view on the B-mode image. This is because CDU suffers from
bleed into the vessel wall or dropout adjacent to the vessel wall and so is not as accu-
rate as measuring from a sharply imaged wall on B-mode.
As an example of combining measurements involving B-mode and Doppler wave-
forms we look at the measurement of volume flow.
MEASUREMENT OF VOLUME FLOW Q
Volume flow is potentially a very useful measurement to make as it is a measure of

the oxygen-carrying capability of a vessel (Figure 10.20). However, its use is limited
due to the way errors accumulate in its measurement and the assumptions that must
be made. [7]
Cross-Section Area A
For an artery, the assumption is that the cross section is circular. The area A may
therefore be calculated by measuring the diameter of the vessel lumen from

the B-mode image. A longitudinal image is obtained clearly showing the vessel
wall with the intimal layer seen. The lumen diameter D is then measured. Using a
A
Q v
FIGURE 10.20 Definition of volume flow as the volume of a cylinder Q given by the cross-
sectional area A times the average velocity of blood v.
Volume flow Q = v. A where v is the average velocity of blood,
A is the cross-sectional area of the vessel
A C
D
B
FIGURE 10.21 Correct cursor position AB and inaccurate positioning CD for diameter
measurement of vessel.
cross-sectional view to measure the diameter may give an error if the cross-section
is not truly perpendicular to the line of the vessel, such that CD was being measured
(Figure 10.21).
D2
A
4
Any error in D will be squared when calculating the area.
Average Velocity
With the Doppler angle cursor accurately aligned, a waveform is obtained and the
average velocity is estimated from the average velocity function on the scanner. In
order to obtain the most reliable estimate the following must be observed:
• For a Doppler waveform that has reverse flow or reduces to zero for part of the
cardiac cycle, ensure that the average velocity curve accurately maps the flow
as reverse or zero at these points (see average velocity in Chapter 8).
• Ensure the sample volume straddles the vessel so all velocities across the lumen
are included as shown on Figure 10.22. However, this is a limited measurement
Range
gate
FIGURE 10.22 The correct setting of the range gate (a) and (b) showing the problem of the focus
on the Doppler beam catching the centre line velocities and missing the low velocities at the lateral
vessel walls.
because the Doppler ultrasound beam is focused at the sample volume depth.
Even with the sample volume straddling the vessel, the higher velocities in
the centre of the vessel will be emphasised over the slower velocities near the
vessel walls. Depending on the velocity profile across the lumen, this may lead
to up to 30% over estimate of the average velocity.
• The measurement must be made over a complete number of cardiac cycles to
ensure flow throughout the cardiac cycle is equally represented, for example,
end diastolic to end diastolic over 3–4 cycles.
• The waveform is then obtained and the average velocity measured over a
complete number of cardiac cycles, for example, 3 cycles end diastole to end
diastole as shown in Figure 10.23.
• Using the fact that measurement error reduces as n, the estimate of volume
flow may be improved by repeating the measurements and averaging the result.
To give an idea of the precision of volume flow measurement, in a brachial artery

diameter 6 mm, the volume flow Q may be estimated as Q ±12%. This is good enough
when considering the very high flows seen in arteriovenous fistula used for hae-
modialysis.
Better estimates of volume flow are possible using ultrafast ultrasound techniques
(see Chapter 13).
FIGURE 10.23 An example of volume flow measurement in a renal arteriovenous fistula

showing diameter measurement perpendicular to vessel walls, and average velocity measured over
a complete number of cardiac cycles.
REFERENCES
1. Public Health England (2021). https://www.gov.uk/guidance/fetal-anomaly-screening-programme-
overview (accessed August 2022).
2. ISUOG (2019). Practice guidelines: ultrasound assessment of fetal biometry and growth.
Ultrasound in Obstetrics & Gynecology 53: 715–723. https://doi.org/10.1002/uog.20272.
3. Benson, R.A., Meecham, L., Fisher, O. et al. (2018). Ultrasound screening for abdominal aor-
tic aneurysm: current practice, challenges and controversies. The British Journal of Radiology 91
(1090): 20170306. https://doi.org/10.1259/bjr.20170306.
4. Kastelein, J.J. and de Groot, E. (2008). Ultrasound imaging techniques for the evaluation
of cardiovascular therapies. European Heart Journal 29: 849–858. https://doi.org/10.1093/
eurheartj/ehn070.
5. Dicuio, D., Pomara, G., Menchini Fabris, F.M. et al. (2005). Measurements of urinary bladder
volume: comparison of five ultrasound calculation methods in volunteers. Archivio Italiano di
Urologia, Andrologia 77: 60–62.
6. Oates, C. (2001). Cardiovascular Haemodynamics and Doppler Waveforms Explained. Cambridge:
Cambridge University Press.
7. Oates, C.P., Williams, E.D., and McHugh, M.I. (1990). The use of a Diasonics DRF400 duplex
ultrasound scanner to measure volume flow in arterio-venous fistulae in patients undergoing
haemodialysis: an assessment of measurement uncertainties. Ultrasound in Medicine and Biology
16: 571–579.
CHAPTER 11
Safety and Quality Assurance
As with anything we do with patients, in performing diagnostic ultrasound scans, we

must ensure that we operate in a safe manner that recognises all the potential hazards
and minimises the risks. A hazard is anything that can cause harm and risk is the
chance that any hazard will actually cause somebody harm. An additional factor to
be considered in the use of a modality such as ultrasound is the benefit of performing
the examination to the risk it incurs. Potential hazards may range from severe (e.g.
injury to an unborn child) to very minor (e.g. clothing is soiled by the ultrasound gel
used). In the case of therapeutic or surgical use of ultrasound, the specific aim is to
cause a change to the tissues being insonated. For diagnostic usage, such changes
must be avoided.
Likewise, the risk may be high or low, so a procedure that has a high risk of a
severe hazard is not one we would routinely perform! The risk versus benefit is also
an important consideration and, for example, comes into the debate about whether it
is right to perform ultrasound solely for souvenir scans.
It is with these considerations in mind that we approach our discussion of the
safety of diagnostic ultrasound.
The first thing to note is that in exposing the body to ultrasound we are putting
energy into tissue. As we saw in the second chapter, apart from the small proportion
of energy reflected back to the transducer as echoes, the rest is absorbed by the
tissue as specular reflection and scattering take place. In fact, at the low megahertz
frequencies used in diagnostic ultrasound the proportion of energy directly absorbed
by tissue is much higher than the proportion scattered, which is less than 10%. That
which is scattered is also eventually absorbed. The deposition of energy will, in gen-
eral, lead to tissue heating. In addition, since the energy is carried in the form of

225
226 Chapter 11 Safety and Quality Assurance
mechanical vibrations of molecules as they experience the sound wave, we may antic-
ipate that there will be mechanical effects within the tissue that may be hazardous.
Throughout this section reference has been made to The Safe Use of Ultrasound in
Medical Diagnosis [1].
ENERGY, POWER, AND INTENSITY
As we saw in Chapter 1, the energy in a sound wave is proportional to the acoustic

pressure p squared.
E p2 measured in Joule J
Power is the rate at which energy is transferred
E
W measured in Js 1 or Watts W
t
Intensity is the power per unit area, or the energy flowing across an imaginary
surface cutting across the ultrasound beam. The higher the intensity, the greater the
energy transferred to that point in tissue (Figure 11.1).
W 2
I for diagnostic ultrasound this is measured in mWcm
area
Intensity is also proportional to the acoustic pressure p squared. For a plane wave,
the instantaneous intensity is given by
p2
I where Z c, the acoustic impedance
Z
So, at regions of high intensity the pressure amplitude p of the sound wave will be
greatest, and hence particle displacement and mechanical stress will be relatively large.
Increasing intensity
FIGURE 11.1 Diagram illustrating the effect of increasing intensity at a point in tissue.
Intensity 227
Intensity is therefore the key measure in relation to consideration of any hazard

caused by insonating tissue.
MEASURING INTENSITY
Acoustic pressure can be measured with a hydrophone. This is a very small cali-
brated ultrasound transducer of known area (e.g. 0.5 mm2) that can pick up the signal
from one point within an ultrasound beam. The amplitude of the signal from the
hydrophone transducer is proportional to the acoustic pressure; hence, the intensity
at that point can be calculated by squaring, or otherwise measuring, the power of the
signal as shown in Figure 11.2.
INTENSITY
In general, diagnostic ultrasound uses pulses of ultrasound to interrogate the target.

Different modalities (e.g. B-mode, CDU, and Doppler) and different machine settings
within each modality will change the pulse shapes and the pulse repetition frequency.
Each pulse delivers energy to the tissue and in between pulses the amplitude is zero.
We therefore need to carefully define what intensity we are measuring.
Within a complex beam shape produced by a transducer, the intensity of the sound
wave varies from point to point, as seen on an intensity contour map (Figure 11.3). At
a given field point within the field of view, insonation is intermittent as it comes in
pulses – both in terms of individual transmitted pulses and ‘pulses’ at the field point
P as the beam sweeps past the field point (Figure 11.4).
+
t
P
–
(a)
t
(b)
FIGURE 11.2 As acoustic pressure amplitude (a) oscillates between positive and negative values,
amplitude squared, the intensity, has a positive value (b) throughout the cycle.
FIGURE 11.3 Schematic intensity contour map of a focused beam. Highest intensity is at
the focus F.
FP
FIGURE 11.4 A field point P experiences a ‘pulse’ of ultrasound as the beam is scanned across
the field of view as the individual pulses are fired in that direction.
In order to assess where the greatest risk of harmful effects might be, we consider
how the intensity varies in ‘space’, i.e. within the volume of tissue insonated, and how
it varies in ‘time’ as ultrasound beams move across the field of view and are only inter-
mittently experienced at a particular field point. A number of defined measurements
are useful (see Figures 11.5 and 11.6).
The spatial peak intensity is the highest intensity found in the transmitted ultra-
sound beam. It is usually at the focus. The spatial average intensity is the average
value across the width of the beam.
The temporal peak intensity is the highest intensity found within a single
pulse, the pulse peak intensity. The temporal average intensity is the intensity aver-
aged across the whole pulse cycle including the dead time between pulses. The pulse
average intensity is the intensity averaged across the pulse length, not including
dead time.
Using these definitions, the following values are defined.
ISPTA spatial peak temporal average intensity. This is the highest value of intensity
in the ultrasound beam averaged over many pulses. In the case of a swept beam
this is averaged over many fields. This measure is the best indicator of potential
heating of tissue by the ultrasound.
Intensity 229
Spatial average Isa
Spatial peak Isp
(a)
I
Spatial peak Isp
Spatial average Isa
(b)
FIGURE 11.5 Definition of spatial peak and spatial average intensity for an ultrasound beam
(a) in plan view of an ultrasound beam and (b) in cross-section.
I
Time average Ita
(a)
Temporal peak Itp
I
Pulse average Ipa
(b)
FIGURE 11.6 Definition of (a) temporal peak intensity and (b) temporal average and pulse
average intensity.
ISPTP spatial peak temporal peak intensity. This is the highest intensity in the
ultrasound beam at the time when it is at maximum, i.e. at the time of peak
pulse amplitude. It is the greatest instantaneous intensity in the field of view.
ISPPA spatial peak pulse average intensity. This is the intensity averaged over the
length of a pulse, not including dead time, at the point in the field of view
where intensity is highest.
ISATA spatial average temporal average intensity. This is the intensity averaged
over the whole beam cross-section and over time at a particular distance from
the transducer. It is the average intensity experienced at a particular field point.
NOTE
The spatial peak intensity will typically be at a transmit focus.
FACTORS AFFECTING DAMAGE POTENTIAL
The potential for ultrasound to damage tissue depends on a number of factors. These
can be summarised as follows:
• Transmit power – increasing transmit power increases the energy input to

the tissue.
• Scan mode (B-mode, Doppler, etc.) – affects pulse shape and pulse length,
pulsing regime used, pulse repetition frequency, and distribution of pulsing
within the field of view. In B-mode the beam is swept across the field of view
and at any one point there will be a significant amount of time when there is
no incident ultrasound, so the average intensity will be low. For Doppler modes
and M-mode, ultrasound is repeatedly transmitted along the same line so the
average intensity will be higher.
• Frequency – attenuation and mechanical effects are frequency dependent.
• Focus – the position and strength of each transmit focus.
• Tissue type – attenuation varies between tissues and sensitivity to injury for
some tissues is of greater significance than others. Particularly sensitive, for
example, are the lens of the eye and early pregnancy when the whole embryo
can be within the sound beam and whole-body development is taking place.
• Dwell times – how long a single target point is within the ultrasound field.
This will be greater for non-scanning modes such as Doppler or M-mode, where
the ultrasound beam is repeatedly fired along the same line, but also depends
on how long the probe is held in one position insonating a particular target.
• Additional factors – for example, the presence of bubbles when a contrast
agent is used.
We will consider these factors in relation to thermal, and mechanical or non-

thermal effects (Figure 11.7).
Thermal Effects 231
Hazard
Thermal Non-thermal
Tissue heating Transducer self-heating Radiation force Cavitation
Cell death Erythema + burning Streaming Stable cavitation Collapse cavitation
Microstreaming Physical damage Free radicals
FIGURE 11.7 The relationship between hazards of ultrasound exposure showing physical
mechanisms and mechanisms of physical and biological damage.
THERMAL EFFECTS
Normal human core temperature is accepted to be 37 °C with a diurnal variation

of ± 0.5–1 °C.
The potential for damage increases with temperature rise and is related to the
length of time an elevated temperature is sustained. In the case of foetal tissue,
increases of T > 2 °C have been shown to produce teratogenic effects in animal
models. Certain periods of foetal development have been shown to be more sensitive
to thermal insult than others, and in general, sensitivity to thermal insult is greater
when tissues contain a large component of actively dividing cells.
The degree of heating within a given tissue will depend on a number of factors.
The main factor is the absorption coefficient of the tissue (see Chapter 2). Generally,
more dense tissues such as bone have a greater absorption coefficient and will there-
fore be heated to a greater extent (Table 11.1). In the case of bone, this may be as much
as 50 times greater than most soft tissue. For soft tissue, absorption is the major com-
ponent of attenuation of ultrasound accounting for some 90% with scattering losses
accounting for no more than 10%.
TABLE 11.1 Attenuation coefficients of different tissues

Tissue Attenuation coefficient α
Blood 0.15
Fat 0.6
Brain 0.8
Liver 0.6
Kidney 1.0
Muscle (along fibres) 1.3 Anisotropy
(Across fibres) 0.7
Water 0.02
Cortical bone 20
Ultrasound
Convection
Conduction
FIGURE 11.8 Mechanisms of energy transport in insonated tissue.
As shown in Figure 11.8, the degree of heating will depend on the balance bet-
ween the energy input by ultrasound and the dissipation of that heat into surrounding
tissue. Dissipation will be by conduction of heat into adjacent tissue and convection
where there is perfusion by blood vessels that can carry the heat away. In terms of
intensity measurement, the key measure for heating is the ISPTA.
In the case of bone, the high coefficient of absorption can lead to significant local-
ised increases in temperature at the bone surface that can raise the temperature of
adjacent soft tissue to levels considered unsafe.
In order to give the user an indication of possible temperature rise from a given
machine setup, the thermal index has been developed and is displayed on the image
display screen.
THERMAL INDEX (TI)
Thermal index is defined as the ratio of the current acoustic output power from the
transducer (W0) to the power required to raise the tissue by 1 °C (Wdeg).
W0
TI
Wdeg
So, a TI of 1 is indicating the possibility that the current output may raise the tem-
perature within tissue by 1 °C.
This is based on a model with a very low level of attenuation, 0.3 dBcm−1MHz−1
(soft tissue = 0.53 dBcm−1MHz−1). In other words, the beam is assumed to be more
intense at depth than it actually is, to give a built-in safety margin.
Because of the possibility of enhanced heating when bone is in the field of view,
three settings of the thermal index have been developed corresponding to different
combinations of soft tissue and bone commonly encountered during scanning.
Thermal Index (TI) 233
TIS
TIS – for soft tissue.
TIB
TIB – for imaging in a non-scanning mode (i.e. with a static beam, for example, pulse
Doppler or M-mode) when bone is at or near the focus.
TIC
TIC – for when bone is near the skin surface, e.g. postnatal cranial scanning.
NOTE
It is important that the correct version of TI is selected for display for the tissue
configuration you are scanning.
NOTES
• TI does not include a dwell time element and for obstetric scanning, guide-
lines shown in Figure 11.9 should be adhered to.
• TI indicates a relative risk rather than an absolute risk. That is, a scan per-
formed using a TI of 2 compared to that using a TI of 1 represents a greater
risk to the patient, although the absolute risk is not known.
Obstetric scanning
Thermal index
0 0.5 1.0 1.5 2.0 2.5 3.0
0.7
Recommended <60 <30 <15 <4 <1 Not
range mins mins mins mins mins recommended
provided for
adequate images OB
can be obtained Recommended scanning time limits for these Tls scanning
(especially in 1st trimester)
(observe ALARA)
Unlimited time
observe ALARA
Monitor TIS up to 10 weeks post-LMP, TIB thereafter.
FIGURE 11.9 BMUS recommended maximum dwell times for increasing TI for obstetric
scanning. Source: Haar [1].
• The TI will change in response to changing the scanner controls.

• TI will tend to be higher in pulsed Doppler mode, where pulses are longer
than for B-mode and the beam is directed continuously at the same target.
This is especially the case for TIB and TIC where bone is included in the
field of view.
• There is likely to be an increased risk to the foetus of any temperature
increase due to insonation by ultrasound when the maternal core tempera-
ture is already elevated, for example, due to maternal fever.
IMPORTANT
Note where on the display the current value of TI is shown and ensure you are
operating within safe limits for the target of interest.
TRANSDUCER SELF-HEATING
When a transducer transmits acoustic energy, the transfer of electrical energy into
transmitted ultrasound is not 100% efficient and the transducer will itself heat up. This
is known as transducer self-heating. It arises from both electrical resistive heating
in the transducer circuitry and from absorption within the transducer of the acoustic
energy generated. Matching between the transducer and the external medium is not
perfect and there remains a significant impedance mismatch between the transducer
and the external medium (see Chapter 4). In the case of a transducer running in air,
i.e. not in contact with the body, almost 100% of the ultrasound generated is reflected
back into the transducer at the air interface. If the temperature of the front surface
of the transducer becomes elevated, there is a risk to the patient when the probe is
placed on the skin.
It is often transducer self-heating that restricts the acoustic power output of a
transducer rather than a limit on ISPTA or MI (see below).
The manufacturers safety standard, IEC60601-2-37(2007) [2] limits the trans-
ducer surface to <50 °C when running in air and to <43 °C when in contact with
the patient.
NOTE
It is important to be aware and ensure that the probe surface, whether an external
probe or an endo-probe, is not getting too hot (43 °C would be a hot shower, 50 °C
will feel burning and too hot for a handwash).
Streaming 235
NONTHERMAL EFFECTS
The causes of nonthermal effects of ultrasound in tissue have a number of mech-

anisms that may be divided into two groups - non-cavitational and cavitational
mechanisms.
RADIATION FORCE
As a result of energy being absorbed by tissue, the tissue experiences a force called
radiation force in the direction of propagation of the ultrasound. The time average
value of this force per unit volume of tissue FV is given by
2 I
FV
c
where α is the absorption coefficient, I the intensity of ultrasound and c the

speed of sound.
The total force FT exerted on the tissue depends on the total power W absorbed
from the ultrasound beam.
W
FT where c is speed of sound
c
Measurement of this force in water may be used to determine the acoustic power
output of a transducer. For example, using a radiation force balance with a total
acoustic absorber, the radiation force from incident ultrasound is balanced against a
weight under gravity to determine the acoustic power (Figure 11.10).
STREAMING
When ultrasound passes through a liquid, the molecules and any suspended particles
are free to move and radiation force will produce the phenomena of streaming. This
US
Total absorber g
FIGURE 11.10 Diagrammatic illustration of a radiation force balance.

(a)
cm
1
2
3
4
5
–16 –12 –8 –4 0
(b)
FIGURE 11.11 (a) Illustration of the movement within a fluid generated by radiation force from
an ultrasound beam. (b) Shows an M-mode scan of bubbles in a beaker full of water. The bubbles
are pushed down by the radiation force from the linear array transducer.
may sometimes be seen within cysts and other fluid-filled cavities in the body, show-
ing up as motion on a CDU display (Figure 11.11).
In normal scanning, radiation force does not present a significant hazard but it is
the force used to probe tissue in some modes of elastography (see Chapter 14).
CAVITATION
An acoustic sound wave is a pressure wave that alternately compresses and stretches
the molecular structure through which it passes. Where there are small (micro-
scopic) gas bodies or bubbles within the medium, they will oscillate in the sound
field. At low intensities they will continue to oscillate and the phenomena is known
as stable cavitation or non-inertial cavitation. Around the oscillating bubble
micro-streaming can occur subjecting cells and sub-cellular structures to high and
potentially damaging velocity gradients as the fluid slows down away from the
bubble (Figure 11.12).
As the amplitude of the sound wave increases (higher intensities) the bubbles
can grow by a process called rectified diffusion and then catastrophically collapse
within the space of one cycle of the ultrasound wave due to the surrounding tissue
Cavitation 237
FIGURE 11.12 Two patterns of microstreaming that may occur around microbubbles.
Source: Adapted from Jalal and Leong [3].
Stable cavitation
Collapse
Bubble growth
cavitation
FIGURE 11.13 Illustration of bubble growth followed by collapse in an oscillating sound field.
bearing down on it as the pressure increases as shown in Figure 11.13. This is called
inertial or collapse cavitation and is the main non-thermal damage mechanism
in tissue.
Rectified Diffusion
Microbubbles are able to grow in an oscillating sound field by a process known

as rectified diffusion (Figure 11.14). The bubble will shrink and expand as the
acoustic pressure changes. When the pressure is low – a phase known as rare-
faction – the bubble expands and has a larger surface area through which gas
dissolved in the surrounding tissue can diffuse. When pressure grows during the
compression phase of the cycle gas in the bubble is squeezed and will diffuse out
of the bubble back into the surrounding tissue. However, in that phase the bub-
ble has a smaller surface area, so the amount of gas leaving during compression
is less than the gas entering during rarefaction. The net result is that with each
compression-rarefaction cycle the bubble grows.
(continued)
(continued)
+
Compression
P
Rarefaction
–
– +
FIGURE 11.14 The movement of gas vapour into and out of a microbubble oscillating in the
sound field.
As a bubble collapses the pressure and temperature within the bubble increases
dramatically and can briefly reach extremely high values (up to 2000 atm and 5000 °C)
in a highly localised space. This causes large mechanical stresses on the surround-
ing medium and potentially creates free radicals that can cause chemical damage.
Free radicals are highly reactive molecules created by stripping off electrons from
stable molecules. The potential damage to tissue includes disruption of cell mem-
branes, haemorrhage of small vessels, and chemical damage to important molecules
such as DNA.
Whether cavitation is possible or not depends on a number of factors. Firstly, it is
more likely if there are existing free gas bodies within the tissue that can act as cav-
itation nuclei on which the cavitation mechanism can commence. Some tissue has
more such cavitation nuclei than others, especially if it is adjacent to gas-filled spaces,
for example the lung and the gut.
The pulse shape and frequency of the ultrasound will affect the likelihood of
cavitation. Whereas thermal effects depend on the average energy input to the tissue,
cavitation depends on the amplitude and length of individual pulses. In particular, it is
the magnitude of the negative pressure experienced during the rarefaction phase of
the sound wave cycle that determines the likelihood of collapse cavitation (Figure 11.15).
It is the transition from a large negative pressure to a positive pressure that causes the
catastrophic collapse of the bubble as the surrounding medium pushes in on it.
Peak compression pressure

+
–
Peak rarefaction pressure
FIGURE 11.15 Definition of peak compression and rarefaction within an ultrasound pulse.
Alara 239
MECHANICAL INDEX (MI)
Corresponding to the thermal index shown on the ultrasound image display, a

mechanical index has been developed to indicate the potential for collapse cavita-
tion to occur in-vivo. It is based on measurements in water that have been de-rated
assuming an attenuation coefficient of 0.3 dBcm−1MHz−1. As this is less than that of
soft tissue (0.53 dBcm−1MHz−1) it will overestimate the likelihood of cavitation occur-
ring and so has a built-in safety margin.
pr,3
MI
fc
where pr,3 is the peak rarefaction (p−) pressure in water assuming an attenuation
coefficient of 0.3 dBcm−1MHz−1 and fc is the centre frequency of the ultra-
sound pulses.
NOTES
• MI is given without specifying units, i.e. it is a numerical index.
• The FDA currently limits scanner output to a maximum MI of 1.9.
• Below an MI of about 0.4 bubble growth is unlikely even if gas bodies are
present in the tissue.
• Lower frequencies will give higher values for MI.
IMPORTANT
Note where on the display the current value of MI is shown and ensure you are
operating within safe limits for the target of interest.
ALARA
As Low As Reasonably Achievable

The ALARA principle is: to work so that any risks are as low as reasonably
achievable.
• Use the minimum transmit power that will give the images you need.
• Reduce dwell time, especially to critical organs, to that necessary to obtain the
information required.
CONTRAST AGENTS
Contrast agents (CA) used in contrast enhanced ultrasound (CEUS) consist of

injectable liquids with suspended gas bodies (see Chapter 12). It is the suspended
gas bodies that are highly reflective to ultrasound, enabling them to show up in low
concentrations and when visualisation is otherwise poor. However, the use of such
contrast agents introduces two distinct areas of risk for harm.
• The pharmacological risk associated with the use of an injectable substance.

• The risk of physical effects arising from the introduction of gas bodies into
tissue that is insonated.
Clinical Risk
Contrast agents are not usually intended to be pharmacologically active but, as with
any substance injected into the body, reaction such as anaphylaxis is a possibility. Con-
trast agents should always be used in accordance with the manufacturer’s information
on the agent and its administration, and knowledge of the patient.
Bio-effects from Ultrasound

With the administration of the contrast agent, gas bodies (bubbles) are introduced
into the tissue being insonated. The threshold intensity for cavitation in soft tissue is
generally very high but the presence of CA bubbles makes the potential risk of cavi-
tation much greater.
The potential bio-effects of CA use depends on the following factors:
• Rarefactional pressure amplitude (P−).

• Agent dose – multiple doses of CA will increase risk and the bio-effects seen.
• Agent delivery – for example, whether the CA is given as a bolus injection or
more slowly.
• Imaging mode – affects pulse shape and amplitude
• Tissue properties – some tissues are more sensitive to cavitational damage
than others (e.g. kidney)
• Clearing pulses (flashing) – For some investigations a ‘clearing pulse’ may
be used that deliberately bursts the bubbles via collapse cavitation so as to look
at, for example, refill times in perfusion studies. Such a clearing pulse has a
high amplitude with a large rarefactional pressure amplitude that is deliber-
ately designed to destroy the bubbles.
THEREFORE NOTE
When using CA the mechanical index provides little dosimetric guidance
for the safe use of CA aside from a general indication of exposure risk. Such risk
therefore needs to be mitigated where possible.
Strategies for reducing risk include:
• For normal and CEUS imaging keep MI to less than 0.4. For MI > 0.7 there is a
risk of cavitation with CA.
• Follow the CA manufacturers guidance and warnings on use and dosage of CA
• Unless required for clinical investigation, avoid exposure of tissue that is partic-
ularly sensitive to cavitational damage, e.g. kidney.
QUALITY ASSURANCE AND ROUTINE CHECKS
Quality assurance (QA) is the name given to the practice of performing a set of rou-
tine checks to ensure that the ultrasound machine is performing to its design specifi-
cation over the lifetime of the machine.
The importance of adequate QA is seen if one considers the problem of a reduction
in quality.
• Poor images – loss of diagnostic information.

• Tendency to increase transmit power, lengthen examinations, etc. to compensate.
• Bio-safety – for example, probe damage resulting from damage during needle
procedures or unclean probes.
For major QA checks and assessing problems, local medical physics or electronics
departments, or the manufacturers of the scanner should be consulted. In particular
they will be able to check that all is as it should be on acceptance of a new machine,
when there are major upgrades and annually thereafter. Records of QA measurements
should be recorded and kept. For more routine checks, the user needs to be respon-
sible. They see the machine every day or week and they will be the first to notice any
change in quality [4].
SUGGESTED ROUTINE USER CHECKS
Every Scan and and probe hygiene is vitally important, especially when using
H
endo-probes. Ensure probe and keyboard remain clean before and
after each scan.
Every Day isually inspect probes for wear and tear and mechanical damage
V
including needle damage on the face of the transducer. Check the
monitor screen is clean and is correctly set up for brightness and con-
trast in normal viewing conditions.
Check the air reverberation pattern for any shadows or streaks in
the image of each probe you use. This is simply an ultrasound image
(a) (b)
(c)
FIGURE 11.16 Method for detecting element dropout in an ultrasound array using a wire tool
(a) and (b). An example showing the reduction in amplitude of the reverberation as the tool passes
a non-functioning element (c). Source: Image Courtesy of Barry Ward.
of the probe in air with no gel and may indicate element drop-out and
changes in the integrity of the front face of the probe (see below).
Weekly isually inspect probes, scanner controls and cables and brake
V
function for wear and tear and mechanical damage.
Check for element dropout for each probe (Figure 11.16). This is
a significant cause of image quality deterioration. It can be simply
checked for, by running a thin metal wire tool (1 mm diameter),
that can be made for the job, along the dry surface of the probe (see
Figure 11.16b). The image of the wire should be uniformly seen
across the probe and dropouts will be apparent as darker lines or
absent echoes on the image.
Monthly Check and clean air filters as needed.

(a)
(b)
FIGURE 11.17 Image of the air reverberation pattern of a linear array probe with images (a)
and (b) taken three months apart showing deterioration in the laminate on the front of the probe.
Source: Images Courtesy of Barry Ward.
With standard control settings, use an air reverberation test, as described

above, to check the sensitivity of each probe and record the image. The image
should show a set of uniformly bright reverberation lines parallel to the front
face of the probe. By recording the image, changes over time can be compared
to spot changes that may indicate deterioration in image quality as shown in
Figure 11.17.
Annual Service ocal medical physics or electronic department, or

L
manufacturer.
THE USE OF TEST OBJECTS
More in-depth QA checks may be performed using test objects and in-vitro phantoms.
These often consist of sets of wires or tubes of different diameters arranged in sets
of patterns that allow axial and lateral resolution to be measured and measurement
caliper calibration to be checked. They may be filled with a water-alcohol mixture
to give a speed of sound of 1540 ms−1. These may be open top test objects (OTTO)
enabling the transducer to be positioned in the fluid (Figure 11.18). Alternatively, they
may be filled with a tissue mimicking gel. The gel has scatterers embedded to mimic
(a)
(b)
FIGURE 11.18 (a) Illustration of an open top test object (OTTO) consisting of wires accurately
spaced in 1540 ms−1 fluid. (b) An image from the OTTO with measurement calipers shown.
Personal Risk Management 245
FIGURE 11.19 Image from a tissue mimicking phantom showing wires, cysts and solid lesions
of different sizes and echogenicity.
the acoustic appearance of soft tissue. Tissue mimicking phantoms often have cylin-
ders of higher and lower echogenicity and of varying diameter embedded in a mean
echogenicity gel to allow contrast resolution and the detection of small lesions to be
assessed as shown in Figure 11.19.
PERSONAL RISK MANAGEMENT
As well as ensuring safe practices regarding the use of ultrasound on the patient
are maintained, it is also important to ensure that risks to you as a sonographer are
properly managed. Work-related repetitive strain injury (WRRSI) is a potential
hazard the sonographer faces [5]. The problem arises from the fact that scanning often
involves micro-manipulation of the probe with the arm extended whilst at the same
time turning to look at the screen and stretching the other arm out to operate the con-
trols. For some types of scan, for example, vascular scans, TV scans and cardiac scans,
the operator may need to adopt an awkward position in order to obtain the images
needed. The problem with manipulating the probe with an extended arm is that all
the muscles of the arm are in tension in order to produce a stable hand, that must then
make very fine, well controlled, movements to visualise the various image planes.
In addition to operating the machine, the sonographer may have to perform some
patient handling.
WRRSI Risk Mitigation

• Ensure couch height, machine height and chair height are all adjusted for com-
fort and ease of scanning – and try changing these for some variation.
• Position the screen so it is viewable with as little head twist or tilt as possible.
• Ensure the positioning of the patient is optimal for the scan being performed.
All patient handling should be in accordance with good practice.
• Where possible rest your scanning arm whilst scanning, for example, on the
patient where appropriate, on the couch or on a chair arm rest, to ensure the
arm is as relaxed as possible. Frequently assess your posture for bad habits.
• Ensure you are holding the probe with an ergonomically efficient grip that does
not stress or strain muscles and joints [6].
• Ensure background lighting in the room is appropriate without any reflected
flare on the image screen (or flare across glasses if you wear them).
• During a scan take micro-breaks – that is, for a couple of seconds between
views, consciously relax your arm and shoulders. Take your hand off the key-
board when not operating the controls.
• Between patients consciously perform stretch and relax exercises of your arm,
shoulder, neck, etc.
• Manage your workload. Too many patients in a session increases the risks.
Time between patients to write the report can be a time when a different set of
muscles are used – ensure desk height, screen height and chair height for typ-
ing reports is appropriate.
• To use different muscle groups try scanning left-handed (or vice versa).
In Summary
• Find what works for you.
• Keep exercised and relaxed.
• Keep it under review.
• Share ideas with colleagues.
• With good risk management, problems should not occur and any risk be
mitigated.
NEW TECHNIQUES IN ULTRASOUND
Finally, when considering quality related to the use of ultrasound, it is easy to over-
look the question of the techniques and measurements we are making on patients.
When a new modality or technical improvement in ultrasound is introduced, such as
adaptive filtering or harmonic imaging, does it actually enhance the imaging we per-
form? Does it improve or at least maintain the quality of our work in using ultrasound
diagnostically? This is perhaps an area of quality that is not often considered and it is
easy to accept changes because they are presented to us by manufacturers. All of the
techniques described in this book have been introduced, or are being introduced, as
improvements. It is something to think about. One attempt to assess a new technique
in a formal way is described in Hemmsen, M.C. et al. [7]. It uses the introduction of
References 247
synthetic aperture imaging, described here in Chapter 13, as an example of moving

from concept through to clinical assessment to ensure clinical value of the end prod-
uct. Another looks at side by side comparisons of new beamforming and adaptive
filtering techniques [8].
REFERENCES
1. Ter Haar, G. (ed.) (2012). The Safe Use of Ultrasound in Medical Diagnosis, 3e. London: British
Institute of Radiology.
2. IEC 60601-2-37:2007+AMD1:2015 CSV (2015). Medical electrical equipment – Part 2-37:
Particular requirements for the basic safety and essential performance of ultrasonic medical
diagnostic and monitoring equipment. https://webstore.iec.ch/publication/22634 (accessed
August 2022).
3. Jalal, J. and Leong, T.S.H. (2018). Microstreaming and its role in applications: a mini-review.
Fluids 3: 93. https://doi.org/10.3390/fluids3040093.
4. Dudley, N., Russell, S., Ward, B., and Hoskins, P. (2014). BMUS guidelines for the regular quality
assurance testing of ultrasound scanners by sonographers: BMUS QA working party. Ultrasound
22: 8–14. https://doi.org/10.1177/1742271X13511805.
5. Murphey, S. (2017). Work related musculoskeletal disorders in sonography. Journal of Diagnostic
Medical Sonography 33: 356–369. https://doi.org/10.1177/8756479317726767.
6. Roll, S.C., Selhorst, B.S., and Evans, K.D. (2014). Contribution of positioning to work-related
musculoskeletal discomfort in diagnostic medical sonographers. Work 47: 253–260.
7. Hemmsen, M.C., Lange, T., Brandt, A.H. et al. (2017). A methodology for anatomic ultrasound
image diagnostic quality assessment. IEEE Transactions on Ultrasonics, Ferroelectrics, and
Frequency Control 64: 206–217. https://doi.org/10.1109/TUFFC.2016.2639071.
8. Hasegawa, H. (2021). Advances in ultrasonography: image formation and quality assessment.
Journal of Medical Ultrasonics 48: 377–389. https://doi.org/10.1007/s10396-021-01140-z.
CHAPTER 12
Advanced Topics
This chapter covers the following advanced topics:
• Contrast Agents
• B-flow Blood Vessel Imaging
• Doppler Measurement of Pressure Gradients
• Advanced Image Processing – Computer Aided Diagnosis
• Fusion Imaging
• Needle Visualisation and Guidance
CONTRAST AGENTS (CA)
For contrast enhanced ultrasound (CEUS) an agent is introduced into the blood-
stream that will show up in contrast to the average B-mode speckle grey level. Blood
normally shows as black or has very low-level echoes as it is a very poor echogenic
target. Gas in the form of bubbles forms a highly echogenic target which when
injected intravenously enables the blood space to be visualised right down to the
micro-vessel level.
In its simplest form, gas bubbles may be introduced by simply injecting intrave-
nous saline that has been shaken up. This is sufficient to demonstrate a patent defect
in the heart as the strong signal of microscopic bubbles in the saline may be detected
in the left carotid artery.

249
250 Chapter 12 Advanced Topics
For contrast studies in general, a contrast medium of encapsulated gas micro-

bubbles suspended in a carrier fluid is used [1]. The encapsulation stops the bubbles
dissolving and prevents agglomeration of the bubbles.
The structure of a contrast bubble is an encapsulating shell enclosing a gas. Exam-
ples of the materials used are shown below.
Gas Shell
• air • albumin
• perfluorocarbon • galactose
• nitrogen • lipid polymers
NOTES
• Micro-bubbles are produced with a fairly uniform diameter of 1–4 μm. This
may be compared with red blood cells ~8 × 2 μm (see Figures 12.1 and 12.2).
• Micro-bubbles are small enough to cross the capillary bed, including the
lung, so remain in full circulation. They may last three to eight minutes.
• Heavier gases such as perfluorocarbon are less soluble so last longer in the
circulation.
• Circulation around the body takes about one minute with the body at
rest. Bubble loss occurs through filtration by the lungs, destruction by the
immune system, passive dissolution, and shear forces in the circulation.
FIGURE 12.1 Perfluropropane bubbles with a protein shell (Optison; GE Healthcare,

Milwaukee, Wis), seen here against a background of red blood cells Source: Wilson and
Burns [2]/Radiological Society of North America.
Behaviour of Bubbles in the Ultrasound Field 251
Gas
Shell 2 μm
4 μm 8 μm 1 μm
FIGURE 12.2 Comparison of size of a typical contrast microbubble and a red blood cell
drawn to the same scale.
• Micro-bubbles are large enough to not cross the vascular endothelium. This
means they stay within the vascular space – unlike CT or MRI contrast
agents which quickly migrate out of the vascular space.
• Leaks from the vascular space may therefore be detected, for example endo-
vascular leaks in aortic stents.
• As bubbles stay within the vascular space and they are a strongly reflecting
agent, typically, only 1–2 mL of contrast is needed.
BEHAVIOUR OF BUBBLES IN THE ULTRASOUND FIELD
Bubbles, including the encapsulated bubbles used as contrast agents, will oscillate in
the sound field as the acoustic pressure changes. At low pressure, they expand and at
higher pressure they are compressed (Figure 12.3).
The relative change in diameter will depend on the peak-to-peak pressure change
in the sound wave and the stiffness of the bubble shell.
• The more elastic the shell wall, the greater the diameter excursion.
• Ambient pressure change such as systolic- diastolic BP, will affect
bubble diameter.
As bubbles oscillate in the sound field, they are not simply passive reflectors of the
ultrasound wave but they re-radiate the sound wave as acoustic sources themselves.
FIGURE 12.3 Change in microbubble diameter with pressure change in the ultrasound wave.
0.01
Backscattered
coeff 0.001
(1/cm/Sr)
0.0001
0 2 4 6 8 10 12
Frequency (MHz)
FIGURE 12.4 Backscattered ultrasound signal vs frequency for the experimental contrast agent
BR14 (Bracco Research SA, Geneva, Switzerland). The wide peak around 3 MHz reflects the fact
there is a range of bubble sizes present in the sample. Source: Adapted from de Jong et al. [3].
The strength of the echo signal depends on the magnitude of the oscillation of the
bubbles. This will be greatest when the transmitted ultrasound frequency matches
the resonant frequency of the bubbles. This is similar to pushing a child on a swing.
When the ‘push’ occurs at the natural frequency of the swinging child (pendulum), the
amplitude of the swing increases. The driving force and the swing are then resonant.
It is a convenient property of the bubble size used in contrast media that they
resonate at the megahertz frequencies used for diagnostic ultrasound (Figure 12.4).
The transmit frequency can then be tuned to enhance the echoes received from
the contrast.
CONTRAST AGENT HARMONICS
When insonated with low intensity ultrasound, they resonate linearly with the
transmitted ultrasound frequency. At higher intensities, their mechanical oscilla-
tion becomes non-linear as it is easier for bubbles to expand than to contract. As a
consequence of this non-linear oscillation, they radiate harmonics of the incident
ultrasound frequency. The echo received from the bubbles contains second and third
harmonics (see Chapter 5) and a subharmonic at half the transmitted frequency.
When encapsulated bubbles burst, they release their gas as free bubbles. The gas dis-
solves very quickly, but as it does so the bubbles radiate harmonics very strongly from
the insonating pulses.
The example in Figure 12.5 shows the harmonics generated for a low intensity
1 MHz ultrasound beam 0.3 MPa and a higher intensity beam 0.6 MPa. At low inten-
sity, the scattered sound is mainly at the insonating frequency, but a subharmonic at
half the transmitted frequency and second and third harmonics are also present. At
higher intensities, a series of harmonics are generated in the backscattered sound.
Contrast agents will show up on a normal B-mode image, but the contrast
ratio between the contrast agent and the tissue B-mode image can be significantly
enhanced by using techniques to detect the harmonics arising from the contrast [4].
Contrast Agent Harmonics 253
10
–10
0.6 MPa
Scattered –20
power
(dB) –30
0.3 MPa
–40
–50
0 2 4 6 8 10
Frequency (MHz)
FIGURE 12.5 Backscattered sound from a 1 MHz transducer insonating Quantison contrast
agent at two transmitted powers. The trace at 0.3 MPa shows a subharmonic at half the transmitted
frequency as well as second and third harmonic peaks. Source: Adapted from de Jong et al. [3].
Amplitude Modulation
The first technique uses a sequence of two pulses, where the first pulse T1 has double the
amplitude of the second T2, enables the non-linear signal from bubbles to be detected
On receive, the echoes from the second, half amplitude, pulse T2 are amplified
by a factor of two and then subtracted from the echoes from the first, larger, pulse T1.
Where the pulses have behaved in a linear way in the tissue, the two echoes subtract
to give a net zero signal. However, scattered echoes arriving from the bubbles will be
different for the two different amplitude pulses due to their non-linear behaviour.
Echoes arriving from contrast will therefore have a non-zero signal when the two
pulses are subtracted. By this means, the contrast will be detected and its signal sepa-
rated from that of the surrounding tissue.
T1 A
T2 B
2 ×T2 C
A–C D
Tissue echo Bubble echo
FIGURE 12.6 Amplitude modulation, where T2 is the same as T1 but half the amplitude, is
used to remove background echoes (a) and allow harmonic echoes from the contrast agent to be
displayed (b). The echoes from T2 are doubled to give C and C is added to A, to give D.
T1 A
T2 B
A+B C
Tissue echo Bubble echo
FIGURE 12.7 Pulse inversion, where T2 is the same as T1 but with the phase inverted, is used to
remove background echoes (a) and allow harmonic echoes from the contrast agent to be displayed
(b). The echoes from T1 and T2 are added to give C.
Pulse Inversion
The second way of detecting contrast bubbles is to use the pulse inversion technique
(Figure 12.7).
For this, two similar pulses are transmitted T1 and T2 but for the second T2, the
phase is inverted. Subtracting the echo of the inverted transmit pulse T2 from that
of the non-inverted pulse T1 removes the tissue signal but leaves the harmonic con-
trast signal.
Both of these techniques separate the non-linear harmonic echoes from the tissue
echoes. The contrast signal may then be displayed together with the B-mode tissue
image or, in a similar manner to X-ray subtraction angiography, a contrast only
image may be produced (Figure 12.8). The techniques can be combined and by using
FIGURE 12.8 A harmonic contrast image of a urothelial carcinoma in the bladder with
normal B-mode image shown for comparison. Source: Gupta et al. [5]/Central European Journal
of Urology.
Advanced Micro-Bubble Techniques 255
contrast pulse sequencing tissue movement suppression, further enhancement of

contrast can be achieved. Contrast pulse sequencing is the technique of interrogating
a scan line a number of times using a sequence of different phases and amplitudes
and then combining the echo signals [4].
Ultrasound contrast media may be administered in one of two ways:
• Using a bolus injection and watching perfusion as the contrast fills a vascular space.
• Using a continuous injection so the circulation fills with contrast.
FLASHING
Using a continuous injection opens up the possibility of using contrast for quantitative
measurements. By applying a single large amplitude transmit pulse with a large P-
(negative pressure), known as flashing, the bubbles can be made to burst, with their
gas dissolving in the bloodstream. Following removal of contrast, by flashing the bub-
bles in this way, quantitative measurements of contrast intensity increase during refill
from the circulating contrast agent can be made [6]. This can give useful information
on organ perfusion. Such measurements require careful, reproducible adjustments of
the scanner controls, in particular output power, focal depth, dynamic range, receive
gain and transmit frequency.
NOTES
• Unlike X-ray or MRI contrast techniques, ultrasound contrast is unique in
being able to be removed and then re-gained in this way.
• The technique of flashing to destroy micro-bubbles followed by reperfusion
measurements is particularly suited to the ultrafast ultrasound technique
where low amplitude plane waves preserve the micro-bubbles and measure-
ments at high frame rates can be made (see Chapter 14).
ADVANCED MICRO-BUBBLE TECHNIQUES
Micro-bubbles may be used diagnostically and therapeutically by labelling the

micro-bubble shell so that it binds to a particular cell receptor as shown in Figure 12.9.
It may then be used to carry a therapeutic drug or vector DNA to that site. The labelled
bubbles will bind to the receptor site to highlight the site on the ultrasound image, for
example, areas of inflammation, active thrombi or cancer cells [7].
For therapeutic use, the bubbles, having been delivered to the target of interest,
can be flashed with a large ultrasound pulse so they break and deliver their therapeutic
agent directly to the target.
Microbubble
Ligand
Receptor
Target cell
FIGURE 12.9 Showing how microbubbles may be labelled so as to attach themselves to

receptors on specific targets.
NOTES
• Be aware of the safety issues involved in using contrast agents as discussed
in Chapter 11.
• Contrast agents should always be used in accordance with the manufac-
turer’s advice.
B-FLOW BLOOD VESSEL IMAGING
This technique enables blood flow to be shown on B-mode images with a similar
greyscale appearance to the B-mode tissue image (Figure 12.10).
Echoes from blood form a similar speckle pattern to those of the surrounding
tissue. However, blood is a very poor echogenic target with an echo strength −60 dB
below that of surrounding tissue. On a B-mode image, fast moving blood shows
up as black.
At the low velocities seen in veins and in the cardiac chambers, some echoes are
seen from blood as the red blood cells then stack up to form what are called rouleaux,
which form larger echo targets (Figure 12.11). This is the cause of the so called ‘whisp
of smoke’ sign as seen in Figure 12.12.
One way to enhance the signal from blood would be to increase the transmit
power but this would also increase the tissue signal by a similar amount and would
have safety implications. Therefore, in order to show the echoes from blood at a level
that may be compared with other tissues, it is necessary to isolate the blood signal and
to amplify it relative to the surrounding tissue signal. B-flow achieves this, enabling
blood to be seen on a B-mode image.
SAG ML
LOGIQ
E9
(a)
LOGIQ 0
E9
RIGHT ICA PROX
(b)
FIGURE 12.10 B-flow images showing (a) blood flow in the neonatal brain and (b) a B-flow only
image of the carotid artery with low echogenic plaques [8].
FIGURE 12.11 Image of red blood cells stacking to form rouleaux.

(a) (b)
(c)
FIGURE 12.12 Blood flow across a valve in the popliteal vein showing low echogenicity in fast
flow (a) and increased echogenicity in stationary flow (b). (c) Shows ‘Whisp of smoke’ in a dilated
left atrium. Source: Courtesy of Chris Eggett.
The process of isolating the blood signal uses coded excitation to detect what
is moving in the field of view. The coded pulses are detected and compressed to give
a strong signal from weak targets, equivalent to a short high amplitude pulse (see
Chapter 6).
B-flow uses coded excitation to detect movement in the field of view. The prin-
ciple of how the separation of blood flow signal may be achieved is as follows. Using
a sequence of two transmitted pulses (P1, P2), slowly moving blood may be detected
by comparing the speckle pattern from a small sample volume within the ultrasound
beam. The pairs of pulses are compared using cross-correlation (see Chapter 9).
B-Flow Blood Vessel Imaging 259
P1 P2 P2
Stationary Slow flow Fast flow
FIGURE 12.13 Showing the change in speckle between two successive pulses P1 and P2 seen by
an ultrasound beam with stationary, slow, and fast flow.
B-flow
signal
intensity Completely
uncorrelated
Blood flow
FIGURE 12.14 Relationship between the B-flow signal and the correlation of the speckle as flow
velocity increases.
The difference between P1 and P2 will be zero where the tissue has remained
stationary but there will be a net signal where the speckle pattern has changed, i.e.
moving blood. The faster the blood moves, the greater the net blood signal detected,
up to a maximum value that will occur when all of the blood seen by the first pulse has
left the sample volume by the time the second pulse arrives as shown in F
igures 12.13
and 12.14.
The strength of the blood signal is then displayed within the vessels on a
B-mode image.
NOTES
• B-flow enables the visualisation of blood flow on a B-mode image.
• The signal strength depends on the speed and turbulence of the blood flow.
Turbulent flow decorrelates faster than laminar flow.
• Small – sub-millimetre vessels may be detected.
• As no wall filter is required, the B-flow image shows flow right up to the
vessel wall.
• The level of B-flow brightness to B-mode image brightness can be adjusted

to improve contrast or remove the B- mode altogether so only blood
flow is shown.
• The B-flow signal may be colour coded to increase its contrast against the
B-mode image.
DOPPLER MEASUREMENT OF PRESSURE GRADIENTS
It is useful to be able to measure the pressure drop ΔP across a vascular stenosis or

a stenosed heart valve as this indicates the haemodynamic significance of the ste-
nosis [9]. Doppler ultrasound may be used to estimate the pressure drop by measuring
the velocity of flow proximal to the stenosis vp at peak systole and the highest velocity
found in the stenotic jet vs [10].
Using a simplified version of the Bernoulli equation, the pressure drop is cal-
culated as follows:
1
Ps Pp vs2 vp2 P units Pa pascal or kg m 1s 2
where ρ is the density of blood = 1060 kg·m−1.

Applying a conversion factor of 1 mmHg = 133.3 Pa, the equation can be written
4 vs2 vp2 P in mmHg
where vs ≫ vp, as is often the case when comparing the velocity in a cardiac stenosis
with the velocities in the cardiac chamber, the equation can be further simplified to:
P 4 vs2
An example of pressure measurement across an aortic stenosis is seen in

Figure 12.15.
Where velocity vector data is available for the whole flow across the stenosis, for
either 2D or 3D data, then the pressure drop may be more accurately estimated using
the Navier-Stokes equation, which gives a full description of the flow in 3D [11].
v
P v v
t
v
where ∇P is the change in pressure P along a given vector, is the change in
t
velocity with time along the vector, v is the vector at a point and Δv is the change in
velocity along the vector (see Chapter 13 for velocity vectors). These calculations are
performed by programmed algorithms in a velocity vector processing package.
Artificial Intelligence 261
FIGURE 12.15 High velocities in the left ventricular outflow tract with the pressure drop
across the stenosis indicated. Multiple sample gates are being used to increase the Nyquist limit.
Source: Courtesy of Chris Eggett.
ADVANCED IMAGE PROCESSING
Having obtained some image data there are ways in which the data may be processed to:
• Make what is seen easier to interpret.

• Make automated measurements from the data.
Both of these are being used on ultrasound images. In order to achieve these ends,
the scanner must identify key features in an image. The advent of 3D ultrasound has
enabled the range of what can be done to be greatly extended.
ARTIFICIAL INTELLIGENCE
There are a number of steps involved in such image processing and sophisticated
algorithms are used to automate the process. These methods fall under the general
heading of artificial intelligence (AI). Artificial Intelligence can be broadly defined
as any human like behaviour displayed by a machine or system. Machine learning
is the study of algorithms that can improve automatically through experience and by
the use of data. One subset of machine learning techniques that has proved very effec-
tive in enabling automatic analysis of images and other AI processes is called deep
learning and is discussed further below.
SEGMENTATION
Image segmentation is the process of identifying edges, surfaces, and textures visible
in an image so that the image is divided into regions of interest. This may be achieved
by manually identifying a region or structure, for example by using a cursor, or it
may be automated. Automated methods may require the user to first identify a gen-
eral region or segmentation seed in the image on which the automatic technique
will work.
At its most basic, it may be as simple as using the data from a particular modality,
such as CDU, and displaying blood flow in a more informative way. Or, detecting the
first interface in a clear fluid path, such as detecting the foetal skin surface within
the amniotic sac. This can then be used to produce advanced 3D displays of the
foetus in utero.
EXAMPLES (1–3)
1. Displaying Blood Vessels

The lumen of a blood vessel may be detected by looking for the surface bet-
ween the blood of low echogenicity and the strongly reflecting vessel wall. An
alternative approach is to use Doppler information or the signal from CEUS to
identify the space where there is flow (Figure 12.16).
2. Displaying the Foetus
The skin surface of a foetus in utero may be identified by using first surface
detection of echoes within the liquor in the amniotic sac. Once a surface has
been segmented it can be highlighted in the displayed image.
Using methods developed for video games, a point of illumination
can be chosen and the surface rendered so as to appear the same as if you
saw it in front of you, with a light casting shadows to show facial features
(Figure 12.17). Internal organs may be displayed by making the surface
semi-transparent.
3. Cardiac Valves
In a similar manner to that used for displaying the foetus, cardiac valves may
be rendered to show what one would see during surgery (Figure 12.18). This
allows planning for prosthetic valve insertion. Using 4D data the motion of the
valve may also be viewed.
Flow can be superimposed to show where a regurgitant jet is emanating
from, as shown in Figure 12.19.
At the other end of the scale, segmentation may be used to detect solid lesion
boundaries within tissue parenchyma, identify image orientation, or identify the
boundaries and fit a model to the volumes identified. In the case of echocardiographic
imaging, this can involve identifying frames in a cine sequence and tracking changes
(a)
(b)
FIGURE 12.16 (a) 3D rendering of the carotid bifurcation using the power Doppler signal.
Source: Courtesy of Esaote. (b) 3D rendering of the vasculature of a hepatocellular carcinoma
revealed by CEUS. Source: Illustration by Courtesy of Koninklijke Philips N.V.
across the cardiac cycle. From basic segmentation, we move into the realm of com-
puter aided diagnosis.
COMPUTER-AIDED DIAGNOSIS (CAD)
CAD may be used to identify key images for human analysis/detection or may go
further to automatically provide a diagnosis [13].
Light source Shading applied
relative to
light source
(a)
(b)
FIGURE 12.17 Foetal surface shown using a ‘light source’ to give a 3D rendering. Source: Illustration
by Courtesy of Koninklijke Philips N.V.
AML
AML
PML PML
(a) (b)
FIGURE 12.18 Mitral valve showing the view from (a) the left atrium and (b) the left ventricle.
Source: Drasutiene et al. [12].
LV
Aortic valve
Mitral valve
Regurgitation
Aorta
LA
FIGURE 12.19 Mitral valve regurgitation shown on a 3D rendering of the heart.

Source: Illustration by Courtesy of GE Healthcare.
Pre-processing Segmentation Feature extraction Classification
FIGURE 12.20 Block diagram of the process of computer aided diagnosis.
The method requires the following steps (Figure 12.20):
• Pre-processing – to ensure the data is in the right form for analysis.

• Segmentation – to isolate the regions for analysis.
• Feature extraction – to identify the features of significance and present rele-
vant measurements.
• Classification – use measurements, or otherwise determine the significance
of the features to categorise the information and possibly offer a concluding
result that may be a clinical result.
Segmentation and feature extraction may be closely linked in that the techniques
used to identify what is significant may also be the same as those used to segment the
image. Feature extraction may go further than basic segmentation and highlight fea-
tures within the segmented areas upon which classification can be made [14].
Having segmented the data to isolate the relevant image region, the next step
is to see if it has the features that indicate a significant finding. This has been done
in two ways:
• Direct identification of features.

• Using deep learning to identify features.
Feature Identification
Four feature categories have been used to identify features of interest:
• Texture – e.g. character of lesion surface; local greyscale texture.

• Morphology – shape of lesion, e.g. smoothness of lesion margin; depth to
width ratio, etc.
• Model-based – e.g. statistical analysis of backscattered echoes.
• Descriptor features – direct description of what is seen in the image, e.g.
shape (round, oval, irregular); calcification; shadowing, etc.
Examples
1. Cardiac Quantification
Using a pre-defined model, the segments of the model heart are matched to a
3D volume set of images of a real heart in a series of adjustments that, in four
stages, give a good match (Figure 12.21). The model is fitted across the cardiac
cycle. Using the fitted model calculations of the chamber volumes and other
cardiac functional parameters may be calculated. The modelling and fitting
use an algorithm derived using deep learning. An alternative approach aims
to model the left ventricle directly from a set of ultrasound images [15], again
using a deep learning algorithm.
2. Soft Tissue Segmentation
Using a deep learning algorithm, the segmentation of a soft tissue mass, where
the change in speckle texture is subtle, can be accurately achieved. In this case
for a benign thyroid nodule with degenerative changes (Figure 12.22).
Classification
Having identified the relevant features, the target is classified. For example, as benign
or malignant. This may be done in a number of ways, for example, using a Bayesian
statistical approach (probability of positive) or using a decision tree based on the
features found, etc.
Deep learning has been applied in a number of stages in the CAD process, for
example in the models used to fit to cardiac images for segmentation, in feature
extraction and in classification [17, 18].
Deep Learning
This uses artificial neural networks to learn to recognise the features of interest
in an image. A neural network is an arrangement of computations that in some ways
mimic the way neurons in the brain work. The computations are arranged in layers
such that the outputs from one layer provide the inputs to another layer. The outputs
Knowledge-based identification Patient-specific adaptation
AP4 view
derived from
3D volume
AP3 view
derived from
3D volume
Heat Chamber Regional Regional

localization alignment alignment alignment
(blood-pool border)
FIGURE 12.21 Automated segmentation of the heart using four stages as shown across the
bottom of the image for two views of the heart (top and bottom lines). Observe how the model
is gradually fitted from the left to the right in each line. Source: Illustration by Courtesy of
Koninklijke Philips N.V.
(a) (b) (c)
FIGURE 12.22 Segmentation of a benign thyroid nodule with degenerative changes using a
deep learning algorithm. (a) the ultrasound image, (b) manual segmentation (normal thyroid
blue, nodule red, cyst green) and (c) automated segmentation using the algorithm. Source: Kumar
et al. [16]/IEEE/CC BY 4.0.
× W1
+
× W2 + f(x)
+
× W3
Input layer Hidden layers Output layer

(a) (b)
FIGURE 12.23 Schematic of a neural network (a). Each node receives input from every node
in the previous layer (b). The inputs are multiplied by a weighting factor and then summed
and processed by an activation function to give an output fed to each node on the next layer.
Source: Adapted from Fortunati [19].
from each layer are modified by the status of the inputs to the layer in an analogous
way to neuron outputs being modified by input signals (Figure 12.23). In this way the
system is able to ‘learn’ and produce useful outputs for future input data.
The deep learning method requires a large dataset of known cases to train a
neural network. The example cases are divided into two. One half is used to train
the network with normal and abnormal images, for example. These are known as
ground truth data. The network then automatically extracts the features it deter-
mines, to classify the images. These features may be completely different to any that
a human might suggest, such as those described above. It is to this extent like a black
box – you don’t know what goes on inside but the answer comes out of the output.
Having trained the network, the second half of the examples are then used to test the
ability of the network to correctly classify the data.
DIAGNOSIS WITH CAD
Having performed the automated analysis, the user is presented with some form of
result. This may be in the form of a categorical result (e.g. yes/no) or a probabilistic
result (e.g. a scale or indication of probability). With all automated diagnostic methods
there is the question of robustness, reliability and the potential for false results. The
same is true for a human diagnostician arriving at a result, but where a machine
derived result is presented to you, the onus is on you to assess whether the result can
be relied upon. Firstly, in the immediate situation, for example, has the correct area in
the image been highlighted, does it look likely. And secondly, is there data to support
the particular CAD algorithm’s results in that area of examination. How does it cor-
respond with your own diagnostic opinion of what is seen? As automated analysis
becomes more prevalent, the risk of divorcing active analysis of the ultrasound data
from simply accepting the machine derived result is a possible issue.
Fusion Imaging 269
Examples of CAD
CAD has been used in ultrasound to classify breast lesions, liver fibrosis, and thyroid
nodules. Accuracies are in the range 90–99% and in some instances may outperform
human classification from ultrasound [20–22].
FUSION IMAGING
Fusion imaging is the technique of simultaneously using images of a target of

interest from more than one imaging modality. In particular the use of simultaneous
synchronous navigation of a 3D reference image set, together with live ultrasound
scanning. Such a reference image set may be from CT, MRI, or PET scanning [23].
This enables the sonographer to more easily observe and carry out a procedure
on a target that shows up clearly on a non-live image from another imaging modality
whilst using a live ultrasound scan to direct the procedure, for example, aspiration,
biopsy, or other minimally invasive procedure. It is especially useful for areas of
poor imaging on ultrasound for example low contrast lesions, shadowed areas from
gas or calcification or where there is poor image quality resulting from overlying
fatty tissue.
The benefits of fusion imaging are:
• Increased confidence in identifying what is observed on the ultrasound image.

• Comparison of features as seen in different modalities.
• Saving on more expensive procedures such as surgery.
• Reducing false positives when treating small lesions under ultrasound guidance.
• For use as a training tool to understand the appearance of structures in differ-
ent modalities.
In order to achieve fusion imaging three conditions are required:
• Scaling – The reference dataset and the ultrasound image must have the same
scale so that dimensions on both images match when they are overlaid.
• Registration – The two image planes must be aligned in 3D space so that the
two imaging modalities are showing exactly the same slice through the target
of interest. This involves re-slicing the reference data set to match the ultra-
sound image plane.
• Synchronous movement – The displayed image plane of the reference data
must continuously re-align to match the current image plane displayed by the
ultrasound, in real time.
We therefore need to know the position and orientation of the ultrasound probe at
all times in order to know the current image plane in space. This tracking is achieved
Sensor x
z
Transmitter
FIGURE 12.24 The position and orientation of the probe may be tracked using an
electromagnetic sensor attached to the probe (a). This picks up pulsed magnetic fields from a
transmitter mounted nearby (b). The transmitter and sensor have coils or other magnetic sensors
mounted in three orthogonal planes to give full 3D information (c).
using an electromagnetic (EM) sensor attached to the ultrasound probe. A magnetic

field is generated by currents passed through coils in a field source or transmitter
situated less than half a metre from the patient. A small position sensor is attached
to the ultrasound probe. When the probe is moved within the magnetic field, an
induced electric current is generated allowing the system to recognise the 3D spatial
position and orientation of the ultrasound probe (Figure 12.24).
Registration with the reference data is achieved either by using external
markers placed on the patient’s skin at the time the reference image data was col-
lected or by aligning anatomical markers – vessels, cysts, calcifications, the xiphoid
process, etc. In the case of anatomical markers, alignment is either performed manu-
ally or automatically using intelligent software. Once registration has been achieved,
the two images can be locked together and synchronised in further positioning of
the ultrasound probe. As the scan proceeds, registration may have to be readjusted to
match the reference image to the ultrasound image for the exam to continue.
NOTES
• The fusion images of the two modalities may be displayed side-by-side or
superimposed on one another.
• Ultrasound modes such as CDU, CEUS and elastography may be used to
enhance contrast of the target of interest.
• The ideal track for inserting a biopsy needle, for example, may be planned
in advance of the examination and shown on the images and the progress of
the physical needle tracked and matched to the ideal track.
An example of fusion imaging is shown in Figure 12.25. There are a number of poten-
tial problems with fusion imaging:
(a) (b)
(c) (d)
FIGURE 12.25 An example of fusion imaging of CT and ultrasound in the liver. (a) ultrasound
superimposed on CT, (b) ultrasound, (c) CT and (d) indication of the probe orientation and slice on
the CT. Registration was achieved automatically using anatomical landmarks. Source: Illustration
by Courtesy of Koninklijke Philips N.V.
• Relative movement between the live image and the reference image reducing
registration accuracy, for example due to breathing or probe pressure on the
patient’s skin.
• Lack of anatomical landmarks such as vessels within the field of view.
• Registration error developing as the scan proceeds.
• Interference of the magnetic fields from metallic objects, or placing the field
generator too far away, causing inaccurate information from the sensor of the
current ultrasound image plane.
NEEDLE VISUALISATION AND GUIDANCE
Real time ultrasound has a vital role in guiding needle procedures such as biopsy,
aspiration, administering anaesthetic blocks, placing central lines, foetal surgery, etc.
These require clear visualisation of the position of the needle tip and knowledge of its
direction of travel.
When a needle is introduced into the image plane perpendicular to the
ultrasound beams it will be seen as a bright reflection often with reverberation
artefact, or comet sign, extending below it (see Figure 7.23). However, it is often not
possible to achieve this alignment, especially when trying to reach deeper struc-
tures. In such cases the needle itself may be barely visible in the image. Larger
needles show up better than fine needles and, if bevelled, the needle tip may scatter
the ultrasound so as to show up on the image. A number of techniques have been
used to improve visualisation of needles and enable them to be guided to their
intended target.
Many needle procedures are performed freehand, but needle guides may be fit-
ted as an attachment to the ultrasound probe (Figure 12.26). The line of the needle
may then be displayed on the image as it is travelling along a known path. The target
may be approached using a planned angle/depth route either in a longitudinal plane,
in line with the long axis of the probe, or in a transverse, short axis direction. The
needle guide ensures the planned angle is maintained.
The problem of knowing exactly where the needle tip is, is compounded by sev-
eral possible problems.
• As the needle is pushed through the tissue there may be some flexing of the
needle, resulting in the tip not being exactly on the planned line.
• The needle may appear bent due to refraction artefact (see Figure 7.8).
• The slice thickness of the image is not zero so within the longitudinal
approach, the needle may still be seen in the image but be to one side of
the target, for example, a small blood vessel viewed in longitudinal sec-
tion. Rocking the probe along its long axis may make this situation clear
(Figure 12.27).
• Using the transverse approach, it is easy to pass the needle tip through the
image plane and beyond it. If this occurs, the acoustic shadow from the needle
may be seen.
FIGURE 12.26 Diagram of a longitudinal L and transverse T needle guide on a linear

array probe.
Vessel
Needle
FIGURE 12.27 Illustrating the problem of the needle and target both being within the slice
thickness but not coincident.
NOTES
• Tracking needle progress may be made in real-time on the image by noticing
the disturbance to adjacent tissue structures as the needle pushes through.
• Needles are best seen in the focal zone, so this should be adjusted to ensure
optimal visualisation.
• Larger needles are more easily seen than fine needles.
• Air bubbles in saline may be introduced to show where the needle tip is but
these then have the disadvantage of obscuring the image at greater depth.
A number of more advanced techniques have been used to improve needle

visualisation [24].
Making the needle acoustically reflective. This may be achieved either by coat-
ing the needle with an acoustically reflective coating or by laser etching the surface of
the needle to make it scatter, rather than reflect, the ultrasound.
Active sensing, for example using an electromagnetic sensor or an ultrasound
transmitter at the tip or an ultrasound emitting tip, to enable the tip position and track
to be actively displayed on the ultrasound image [25]. This enables the needle to be
homed in on the target of interest whilst avoiding other sensitive structures.
Image-based automatic needle tracking has been developed to continu-

ously display its position in the image. This may be done by analysing changes in the
B-mode image caused by the presence of the needle or by using power Doppler to
detect needle vibrations.
Using such tracking methods, together with 3D imaging, the exact track of the
needle in space may be visualised.
REFERENCES
1. Stride, E. (2015). Physical principles of microbubbles for ultrasound imaging and therapy.
Frontiers of Neurology and Neuroscience 36: 11–22. https://doi.org/10.1159/00036622.
2. Wilson, S.R. and Burns, P.N. (2010). Microbubble-enhanced us in body imaging: what role?
Radiology 257: 24–39.
3. de Jong, N., Emmer, M., van Wamel, A. et al. (2009). Ultrasonic characterization of ultrasound
contrast agents. Medical & Biological Engineering & Computing 47: 861–873.
4. Whittingham, T.A. (2005). Contrast-specific imaging techniques: technical perspective. In:
Contrast Media in Ultrasonography, Basic Principles and Clinical Applications (ed. E. Quaia),
43–70. Berlin: Springer.
5. Gupta, V.G., Kumar, S., Singh, S.K. et al. (2016). Contrast enhanced ultrasound in urothelial
carcinoma of urinary bladder: an underutilized staging and grading modality. Central European
Journal of Urology 69: 360–365.
6. Tang, M.X., Mulvana, H., Gauthier, T. et al. (2011). Quantitative contrast-enhanced ultrasound
imaging: a review of sources of variability. Interface Focus 1: 520–539. https://doi.org/10.1098/
rsfs.2011.0026.
7. Lee, H., Kim, H., Han, H. et al. (2017). Microbubbles used for contrast enhanced ultrasound
and theragnosis: a review of principles to applications. Biomedical Engineering Letters 7: 59–69.
https://doi.org/10.1007/s13534-017-0016-5.
8. GE White Paper (2012) B-flow Technology, Matt Berger, Qian Adams GE Healthcare 2012.
9. Nguyen, T., Hansen, K.L., Bechsgaard, T. et al. (2019). Non-invasive assessment of intravascular
pressure gradients: a review of current and proposed novel methods. Diagnostics 9: 5. https://
doi.org/10.3390/diagnostics9010005.
10. Harris, P. and Kuppurao, L. (2016). Quantitative Doppler echocardiography. BJA Education 16:
46–52. https://doi.org/10.1093/bjaceaccp/mkv015.
11. Olesen, J.B., Villagómez, H.C.A., Møller, N.D. et al. (2018). Non-invasive estimation of pressure
changes using 2-D vector velocity ultrasound: an experimental study with in-vivo examples.
IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control 65: 709–719. https://doi.
org/10.1109/TUFFC.2018.2808328.
12. Drasutiene, A., Sigita Aidietiene, S., and Diana Zakarkaite, D. (2015). The role of real time three-
dimensional transoesophageal echocardiography inacquired mitral valve disease. Seminars in
Cardiovascular Medicine 21: 16–26. https://sciendo.com/article/10.2478/semcard-2015-0003.
13. Huang, Q., Zhang, F., Li, X. et al. (2018). Machine learning in ultrasound computer-
aided diagnostic systems: a survey. BioMed Research International 5137904. https://doi.
org/10.1155/2018/5137904.
14. Bass, V., Mateos, J., Rosado-Mendez, I.M. et al. (2021). Ultrasound image segmentation meth-
ods: a review. AIP Conference Proceedings 2348: 050018. https://doi.org/10.1063/5.0051110.
References 275
15. Kim, T., Hedayat, M., Vaitkus, V.V. et al. (2021). Automatic segmentation of the left ventricle
in echocardiographic images using convolutional neural networks. Quantitative Imaging in
Medicine and Surgery 11: 1763–1781. https://doi.org/10.21037/qims-20-745.
16. Kumar, V., Webb, J., Gregory, A. et al. (2020). Automated segmentation of thyroid nodule, gland,
and cystic components from ultrasound images using deep learning. IEEE Access 8: https://doi.
org/10.1109/access.2020.2982390.
17. Wang, Z. (2022). Deep learning in medical ultrasound image segmentation: a review.
arXiv:2002.07703v3 [eess.IV] https://doi.org/10.48550/arXiv.2002.07703.
18. Looney, P., Stevenson, G.N., Nicolaides, K.H. et al. (2018). Fully automated, real-time 3D ultra-
sound segmentation to estimate first trimester placental volume using deep learning. Journal of
Clinical Investigation Insight 3: e120178. https://doi.org/10.1172/jci.insight.120178.
19. Fortunati, V. (2019). Quantib. https://www.quantib.com/blog/how-does-deep-learning-work-
in-radiology (accessed August 2022).
20. Kim, S.Y., Choi, Y., Kim, E.K. et al. (2021). Deep learning-based computer-aided diagnosis in
screening breast ultrasound to reduce false-positive diagnoses. Science Reports 11: 395. https://
doi.org/10.1038/s41598-020-79880-0.
21. Gatos, I., Tsantis, S., Spiliopoulos, S. et al. (2016). Quantitative imaging and image process-
ing, a new computer aided diagnosis system for evaluation of chronic liver disease with
ultrasound shear wave elastography imaging. Medical Physics 43: 1428–1436. https://doi.
org/10.1118/1.4942383.
22. Gao, L., Liu, R., Jiang, Y. et al. (2018). Computer-aided system for diagnosing thyroid nod-
ules on ultrasound: a comparison with radiologist-based clinical assessments. Head & Neck 40
(4): 778–783.
23. D’Onofrio, M., Beleù, A., Gaitini, D. et al. (2019). Abdominal applications of ultrasound
fusion imaging technique: liver, kidney, and pancreas. Insights into Imaging 10: 6. https://doi.
org/10.1186/s13244-019-0692-z.
24. Scholten, H.J., Pourtaherian, A., Mihajlovic, N. et al. (2017). Improving needle tip identification
during ultrasound-guided procedures in anaesthetic practice. Anaesthesia 72: 889–904. https://
doi.org/10.1111/anae.13921.
25. Xia, W., Noimark, S., Ourselin, S. et al. (2017). Ultrasonic needle tracking with a fibre-optic
ultrasound transmitter for guidance of minimally invasive fetal surgery. Medical Image
Computing and Computer-Assisted Intervention − MICCAI 10434: 637–645. https://doi.
org/10.1007/978-3-319-66185-8_72.
CHAPTER 13
Ultrafast Ultrasound
Most of the advanced techniques that have been introduced on ultrasound scanners
in recent years are only possible because of advances in computation. In particular
• Speed of the main processors (CPU).

• Memory size and speed of access.
• Graphics processing units (GPUs) capable of massive parallel processing of
image and signal data.
Among other things, these developments in computation have been driven by

the video games industry. It is now possible to acquire, digitise, and store sequences
of complete echo data arriving at all the individual elements of the ultrasound trans-
ducer in real time. By complete echo data, we mean the raw RF data of phase and
amplitude at ultrasound megahertz frequencies (see Chapter 6). This implies digitis-
ing and storing the echo signal at a sample rate of ~40 MHz which quickly generates
giga-bytes of data.
This data can then be further processed either in real-time, or offline, once the
patient has left the examination room. These advances have led to what has been
called ultrafast ultrasound (UFUS) [1].
In ultrafast ultrasound, all of the processing, including beamforming, is performed
by software. It is therefore not limited by having a restricted number of hardware
processing channels.

277
278 Chapter 13 Ultrafast Ultrasound
SYNTHETIC APERTURE IMAGING (SA)
The conventional pulse-echo technique with beam steering and focusing that we have
discussed in previous chapters uses delay-and-sum beamforming. Pulses from
individual elements are delayed so that their phase across the elements used adds
up, forming a beam in the direction and with focus as required. In reception, the
echoes are similarly delayed and their signals summed to give the received signal.
This process is applied to individual pulse-echo sequences to build up the image.
High speed processing enables a novel and conceptually very different way of
image formation, called synthetic aperture imaging, to be performed that brings
lots of additional advantages.
Synthetic aperture imaging is a technique in which an ideal ultrasound source is
synthesised by firing the elements of the transducer so as to recreate the beam that
would have come from the ideal source. In principle, all of the elements of the trans-
ducer are used for each transmission-receive sequence rather than a small group of
elements as used in conventional beam forming. In practice, algorithms that use a
selection of elements, a sparse array, can be used to increase processing speed and
computational load whilst maintaining the resolution.
The ideal source may be a point source situated some distance behind the ultra-
sound probe face as shown in Figure 13.1a. The ‘point source’ can be in any position
to allow wavefronts to be sent out in different directions. Another arrangement is to
use a source that is effectively at an infinite distance behind the array producing plane
waves at the array.
Plane wave beam forming is a common implementation of synthetic aperture
imaging in ultrasound and is the form we will explain here. Diverging beam SA
imaging is used by phased arrays.
FIGURE 13.1 Examples of an array transmitting wavefronts that would have come from ideal
sources. (a) A point source situated behind the array and (b) plane waves from a source situated an
infinite distance behind the array.
FIGURE 13.2 Raindrops falling in a puddle. Note how the curved wavefronts from each drop
position arrive at the edge of the ‘pond’ (array) in sequence along the edge.
In order to visualise how this technique works, it is useful to watch raindrops

falling in puddles or onto a rectangular tray filled with water (Figure 13.2) [2]. Each
raindrop creates a wave that moves out from the point where it fell and in time there
will be a drop on every point in the puddle. The waves move outwards until they reach
the edge, in the same way as echoes returning from the focus of a focused ultrasound
beam reach the aperture of an ultrasound transducer. At the edge, there will be a time
delay across the edge as each part of the expanding wave meets the edge. The wave
from each raindrop point will have a different set of delays in its arrival across the
edge that corresponds to the point where it originated. Keep this picture in mind.
PLANE-WAVE BEAMFORMING
A single plane wave pulse is transmitted by firing all the elements across an array
(Figure 13.3).
On receive, the phase and amplitude of the echoes arriving back at the trans-
ducer are stored for each element in the array, for the entire depth of field of the scan
(Figure 13.4). For a single field point P1, we can take the stored data for each element
and retrospectively shift its relative timing so it corresponds to the time delays of the
echo from that field point. If we do this for all the elements across the array, we rec-
reate the pulse as it would have arrived back from the point P1. Then, since data for
the whole image field is stored, this can be done for every point in the image and the
Δt 2Δt 3Δt 4Δt
FIGURE 13.3 Transmission of a forward and steered plane wave using the whole array.
RF memory
E1 E2 E3 E4 E5
t10
t1
d1 d1
d2 d2
P2
d3 d3
P1
d4 d4
d5 d5
FIGURE 13.4 The RF signal for every element is stored for the whole field of view for each pulse
echo sequence. By applying time delays to the stored data, the wavefronts arriving from any field
point can be retrospectively reconstructed, e.g. P1 and P2.
whole image receive focused. This is just like watching and reconstructing the wave-
fronts from the raindrops from every point in the puddle.
For example, consider a 5 × 10 pixel image. The RF data memory stores the phase and
amplitude of echoes from each element E1, . . ., E5 in the array, arriving at times t1, . . ., t10.
The data in the memories E1, . . ., E5 are combined with focusing delays to focus
on a field point P1 from depth d4 in the body to give a received focus signal from P1.
The same is done for, e.g. field point P2 at depth d3, and similarly for every other
field point corresponding to every point in the image.
If this is done for every scanline (element line) and every depth, then every pixel
in the image will be focused on receive.
By this means, a complete receive-focused field of view is obtained from a single
transmit pulse.
The computational processing is very fast and for a 20 cm deep field of view and a
128 element array, a frame rate of 3.8 kHz can be obtained. This can be compared to
a frame rate of 30 Hz for conventional processing.
By combining the stored data in this way, we can produce an image with receive
focus for every pixel. However, because the transmission pulse was a plane wave and
not focused, image contrast and lateral resolution will be poor. This may be overcome
by transmitting several plane wave pulses at different angles from the transducer
array on successive transmissions, e.g. T1, . . ., T3, and receiving echoes from the same
three steered angles (Figure 13.5) [3].
Having received and stored the RF echo data for each element, from each of
the transmission pulses, the echo data from each pulse can be delayed and summed
with the other transmitted pulses to recreate a transmission focus for every pixel
in the image, in the same way as was done for receive focusing of each image pixel
(Figure 13.6). It is as though we had run a film of the raindrops backwards so that a
set of waves from just three directions converged on each splash point in the puddle.
E1 E2 E3 E4 E5 E1 E2 E3 E4 E5 E1 E2 E3 E4 E5
t10
t
T1 T3 T1 T2 T3
T2 RF memory
FIGURE 13.5 By transmitting a number of angled beams transmit information is stored for
transmit focus reconstruction.
T1 T2 T3
d1
d2
P2
d3
P1
d4
d5
FIGURE 13.6 By applying delays to the stored transmit data for the angled transmitted beams a
transmit focus can be reconstructed for each field point in the image.
1° 5° 20° 40°
20 (a) (b) (c) (d)
20
Depth in mm
30 30
40 40
50 50
10 20 30 40 10 20 30 40 10 20 30 40 10 20 30 40
12 16 24 30
Contrast in dB
Maximum frame rate in Hz
15 000 3 000 750 325
FIGURE 13.7 Relationship between frame rate, number of transmit angles used and image
contrast in plane wave transmit focusing. The effect of increasing number of transmit angles may
be seen in the images (a–d). Source: Bercoff [4]/IntechOpen.
By this means, each pixel in the image can be focused in both transmit and receive
and image contrast and resolution are restored.
Having formed the transmit-receive focused data set, the resultant RF data may
be processed using the equivalent of envelope detection to obtain the amplitude data
for a B-mode image display in the normal way (Chapter 6).
There is a trade- off between frame rate and image quality because several
transmit pulses must be transmitted for each frame. However, like-for-like, there is
still an increase in frame rate over conventional imaging of 5–10 times faster using
plane-wave beamforming. For example, 30 fps conventional becomes 300 fps UFUS
(Figure 13.7).
This technique of reconstructing transmit focal points using several angled
pulses is called coherent spatial compounding (SCS), as the phase of each
pulse is lined up coherently to reconstruct a transmit focus for each point in the
field of view.
NOTES
• Coherent spatial compounding at RF frequencies enhances resolution
by focusing the transmit beam to each field point.
• Incoherent spatial compounding of the displayed image amplitude data,
as used in conventional B-mode compound scans, reduces speckle. This
can also be performed in UFUS by collecting several amplitude detected
images for different reconstructed transmit angles and compounding them
in the normal way.
NOTES
• Lateral resolution is dependent on the maximum angle of the transmit pulses.
• Contrast depends on the number of angles.
• Signal-to-noise ratio (SNR) depends on the number of angles. If the signal-
to-noise ratio is made equal to the signal-to-noise ratio for conventional
focused ultrasound then with plane-wave beam forming, the signal-to-noise
ratio is better across the entire image. In conventional imaging, the SNR is
worse outside the focal zone.
• Axial resolution depends on the length of the pulse.
• Focusing on transmit and receive with plane-wave beamforming, together
with adaptive beam forming can produce diffraction limited resolution
across the entire image – i.e. as good as it can get.
• MI and ISPTA are improved as plane wave transmissions are used with no
focusing creating zones of high intensity.
SUMMARY
• Transmit plane waves using the full width of the ultrasound array.
• Digitally record the phase and amplitude of the received echoes for
every element.
• Reconstruct a receive focus for every image pixel by coherent summation of
element signals.
• Reconstruct a transmit focus for every image pixel by coherent summation
of several angled transmit pulses.
• Final image quality depends on the number of angled transmit pulses used.
• There is a trade-off between image quality and frame rate.
• Like-for-like in image quality, ultrafast ultrasound is 5–10 times faster than
conventional processing, e.g. frame rate 30 becomes 300 frames per second.
• New applications for ultrasound are made possible.
SPEED OF SOUND CORRECTION
Variation in speed of sound leads to general distortion within the image and frank
speed of sound artefacts may be seen. Chapter 7 described how poor image quality
due to speed of sound aberrations may be improved by modifying the system velocity
from 1540 m s−1 to one that more closely matches the speed of sound in the tissue
being imaged. High-speed processing enables full speed of sound correction to be
performed. The principle of these advanced techniques is described. [5]
The average speed of sound in soft tissue is taken to be 1540 m s−1. In reality speed
of sound varies around this value by ± 5% between different tissue types and any beam
path is likely to pass through several tissue types, each with a different speed of sound.
This leads to misplacement of the echoes in the image as shown in Figure 13.8.
The principle behind speed of sound aberration correction is to map the variation
in speed of sound across the field of view and then use the mapped values to correct
the timing of pulses so the echoes for each pixel in the image match their true geomet-
rical location in the image plane.
To understanding how this is done, it is helpful to consider a simple model of a
uniform tissue with a speed of sound C0 = 1540 m s−1 containing a mass with a differ-
ent speed of sound C1 within it (Figure 13.9). If we look at two identical path lengths
to a field point P, AP, and BP, in which one crosses the mass and the other does not,
there will be a difference in time τ of arrival of echoes along the different paths due to
differences in speed of sound along each path.
C0
C1
Distorted
wavefront
FIGURE 13.8 Distortion of the wavefront caused by a tissue with a slower speed of sound C1
than the surrounding tissue C0.
A B
d C1
C0
FIGURE 13.9 Simple tissue model showing two identical length paths for the ultrasound beams
AP and AB with an inclusion with speed of sound C1 in a tissue with speed of sound C0.
1540
For field point P, the ‘go and return’ time of the echo to A is t AP
2d
1540 and C1
and to B is t BP
2d
Because of the tissue mass with speed of sound C1 C0, there will be a time
difference τ = BP − AP in ‘go and return’ times between the two paths.
Where
C1 C 0 will be negative echo BP arrives sooner than AP
C1 C 0 will be positive echo BP arrives later than AP
In the case of real tissue there may be many changes in speed of sound throughout
the pulse path and τ will depend on all the changes along the paths AP or BP. The time
difference τ is the result of all the changes of speed of sound along those beams. This
will be seen as a phase difference between the two signals as shown in Figure 13.10.
By gathering data from many pairs of beam angles, a data set of differences in time
of flight τ along the different paths is obtained from which a map of speed of sound
can be reconstructed using CT algorithms. This is similar to the way as X-ray CT uses
lines of absorption data to produce an X-ray attenuation image. This procedure has
been called computed ultrasound tomography in echo mode (CUTE). [6]
The resulting speed of sound map (Figure 13.11) may then be used in one
of two ways:
• Diagnostically – the image can be displayed as a colour map to show speed of

sound. This gives the sonographer new information that may be diagnostically use-
ful. For example, in breast scanning, lesions may be hard to identify on B-mode but
have a high a speed of sound compared with surrounding tissue, and so show up in
speed of sound mapping. Liver disease may be seen as a change in speed of sound.
• The speed of sound map may be used to correct the time delays used in beam
forming for normal imaging. Using the plane-wave beam forming technique,
when focusing for each pixel is performed, the echo data from each element is
delayed or advanced using the speed of sound map to correct for the exact time
of flight for that echo from that field point. This technique is called local speed
of sound adaptive beamforming. [7]
Δt = τ
BP echo
AP echo
FIGURE 13.10 Phase difference between two echo paths from a field point where C1 < C0.
C0
C1
FIGURE 13.11 Speed of sound map constructed from time differences for different field
point paths.
Figure 13.12 shows an example of speed of sound correction using CUTE. The
images may be compared with Figure 7.9 which shows the effect of speed of sound
aberration from the same phantom with and without compound scanning. Note the
definition in the image of the wires in each case. Figure 13.13 shows a clinical example
of speed of sound correction.
Benefits
Speed of sound correction gives:
• A big improvement in lateral resolution as the reconstructed beams are not

distorted by different, unknown, time-of-flight path lengths within the beam-
width or within any compound scanning.
• Accurate geometrical placement of echoes in the B-mode image, as speed of
sound aberrations have been corrected.
• Improved image quality in both 2-D and 3-D imaging.
• More accurate clinical measurements of dimensions on the image.
• Better matching of ultrasound images to other images in fusion studies.
• Ability to resolve detail at depth in the larger patient.
ULTRAFAST DOPPLER
Ultrafast Colour Doppler (UFCD)

Using the ultrafast technique of transmitting plane waves from multiple angles to
create an image in which every pixel is focused also enables Doppler information to
be gathered at high speed.[9]
0
5 1580
1580 1560
10
1560
15 40
c (m s−1)
z (mm) 1540
20 1520 20
25 1500 1500
1480 1480
30
1460
35 1460
40
0 10 20 30 0 10 20 30
(a) x (mm) (b) x (mm)
0 0
5 5
10 10
15 15
z (mm)
z (mm)
20 20
25 25
30 30
35 35
40 40
0 10 20 30 0 10 20 30
(c) x (mm) (d) x (mm)
FIGURE 13.12 Using a phantom with an inclusion having a different speed of sound (SoS)
(a) shows the SoS map using CUTE, (b) shows the measured SoS across the dashed line,
(c) B-mode image from a single forward angle corrected for SoS and (d) a compound scan
corrected for SoS. Source: (a), (c) and (d) Jaeger et al. [8]/with permission of Elsevier.
0 0
5 5
10 10
15 15
z (mm)
z (mm)
20 20
25 25
30 30
35 35
40 40
0 10 20 30 0 10 20 30
(a) x (mm) (b) x (mm)
FIGURE 13.13 Speed of sound correction in the forearm (a) B-mode image showing blurring of
targets below the fatty layer indicated by the arrow and (b) the same image after correction using
CUTE showing uniform lateral resolution throughout the image. Source: Jaeger et al. [8]/with
permission of Elsevier.
One problem with conventional CDU is that each image line requires ~10 transmit
pulses to sample the Doppler information along a line. The ensemble length is
10 pulses. This requirement for each colour line reduces the frame rate for CDU to
around 10–20 Hz, often restricted to a limited region of the image to improve resolu-
tion and frame rate.[10]
To form a CDU image using plane-wave beam forming, plane wave ultrasound
pulses are transmitted over a small number of angles at the maximum PRF for the
depth of interest (Figure 13.14). In order to get Doppler velocity information that
sequence is repeated 10 times. The Doppler image is therefore acquired much faster
(e.g. 10–20 times) than for conventional CDU.
As for ultrafast B-mode, RF data is collected for every pulse-echo sequence and
stored, e.g. for 5 angles. Sequence T1 to T5 is then repeated 10 times. This gives an
ensemble length of 50 pulses to each field point. All 5N datasets are stored and CDU
detection on the dataset, for example, using autocorrelation, is performed to detect
the mean velocity in every pixel in the whole image. The sign (±) of the Doppler fre-
quency, i.e. whether it is higher or lower than the transmit frequency, determines the
direction for colour coding as with normal CDU (e.g. red-blue).
In addition to the speed of acquisition there are other advantages:
• Velocity resolution is improved to pixel level.

• Every pixel in the image has colour information, not just a localised
colour box.
• There is very good time resolution and all the colour information across the
image is derived simultaneously. This removes the time collection artefact seen
in large CDU box acquisition (see Chapter 10).
• The ensemble length is effectively much higher than for conventional CDU
(e.g. 5 × 10 pulses = 50 pulses). This gives a less noisy estimate of the Doppler
frequency.
• There is increased depth penetration because plane-wave beam forming has
increased sensitivity giving a higher SNR.
• There is high sensitivity to low/small flows.
• By selecting various firing sequences for transmit we can optimise the image
for high or low flows, or for fast changing flows.
T2 T3 T4
T1 T5
FIGURE 13.14 Plane waves are transmitted over a number of angles to obtain Doppler
information.
Ultrafast Doppler 289
Microvascular Imaging
Conventional CDU is not able to visualise vessels <1 mm diameter or with very low
velocity flows. The problem with imaging such vessels with conventional techniques
is that the Doppler shifts from these vessels covers the same frequencies as the motion
of the tissue surrounding them. They are therefore excluded by the high pass clutter
filter CDU uses to avoid tissue motion artefacts.
In order to image these microvessels, a more sophisticated clutter suppression
algorithm must be used. Instead of filtering frequencies in the time domain, it uses
relative changes in the speckle pattern of the solid tissue to the speckle pattern of moving
blood to discriminate between the two [11]. As the tissue with blood vessels embedded
moves, the speckle pattern from a particular point in the tissue moves with the tissue. The
speckle pattern from blood in a vessel will also move position but in addition will have an
additional relative motion to the moving tissue. The scatterers in the flowing blood have
become spatially reorganised to give a new speckle pattern. By applying a spatial filter that
can discriminate between speckle that stays the same whilst it moves and speckle that
changes as well as moving, the blood motion can be pulled out from the general motion of
the tissue and displayed in colour on the image. To detect these small differences requires
the tissue/flows to be sampled many times (colour frame rates of 50–60fps) and it is the
advent of UFUS that has made this possible. Further processing to remove image degra-
dation due to tissue motion may be applied, such as combining sequences of pulses whose
phase and amplitude are changed. This becomes important for looking at very low flows
where longer acquisition times are needed. The end result is that low flows in very small
vessels can be visualised. Examples are shown in Figures 13.15 and 13.16.
M2
4.0 cm
FIGURE 13.15 Microvascular imaging in a renal transplant showing small vessels right out to
the external border of the cortex. Source: Illustration by courtesy of Koninklijke Philips N.V.
(a) (b)
FIGURE 13.16 Vascularity in a ductal carcinoma of the breast. (a) CDU image and (b)
microvascular 3D image. Source: Zhang et al. [12]/Dove Press Ltd./CC BY NC 3.0.
NOTE
Microvessels may be imaged by this technique without the need for contrast
agents although contrast agents can improve visibility and SNR.
Pulsed Wave Doppler (PW)

With UFCD, a PRF as high as that used for conventional pulse Doppler can be achieved
for the whole image. This means that a PW Doppler sonogram can be produced for
any pixel in the UFCD image simultaneously with the CDU display. There is no degra-
dation of the CDU image or sonogram as is the case with such time sharing with con-
ventional CDU. Furthermore, a sonogram may be shown for multiple points within
a blood vessel simultaneously (Figure 13.17). If the UF Doppler data set is obtained
and stored across a complete cardiac cycle, then the CDU map may be retrospectively
shown for any point in the cycle with multiple PW sonograms chosen and shown with
the cycle point marked on them.
NOTES
• Using vector flow imaging, the Doppler angle for the sonogram is automat-
ically determined by the scanner.
• PW Doppler shows much reduced intrinsic spectral broadening than
conventional PW [14].
FIGURE 13.17 Illustrating the principle of showing multiple sonograms simultaneously.

Source: Bercoff et al. [13]/SuperSonic Imagine S.A.
VECTOR FLOW IMAGING (VFI)
In order to completely characterise the flow in the heart or in blood vessels it would
be useful to show the direction and speed of blood flow at every point in the vessel.
This is what the display of flow vectors does. It can yield information on areas of
turbulence, enable volume flow to be measured and enable pressure drops across a
stenosis to be estimated noninvasively [15, 16].
The Flow Vector

To fully specify the motion of a particle of blood we need to know both its speed and
direction of travel. These two together form the flow vector v for that particle. (The
bold ‘v’ indicates that it is a vector rather than a simple velocity or speed.)
We can relate the flow vector of the particle to the axes of the ultrasound image
where X is the transverse axis across the image and Z is the depth in the image away
from the transducer as shown in Figure 13.18.
v.sin θD x
θD
v
v.cos θD
v
z
FIGURE 13.18 The flow vector v at Doppler angle θD showing the x and z components
of velocity.
θD is the angle of the flow from the z-axis and is the familiar Doppler angle. The
length of the vector v shows the speed or magnitude of velocity. The component
of speed vz in the Z direction is given by v.cos θD, which appears in the normal Doppler
equation, and the component of speed in the X direction vx is given by v.sin θD.
In order to determine the vector of a blood particle in the image we need to
find both its velocity in the Z direction and in the X direction. Standard Doppler
measurement finds the velocity in the Z direction as given by the Doppler equation
cfD
vz cos D
2 fT
As shown in Chapter 10, CDU estimates the velocity in the Z direction by

comparing the RF signal from several successive pulses to see how the echo pattern
has moved between them using the method of autocorrelation. What conventional
CDU shows is a map of vz and cos θD remains unknown. If the ultrasound beam was
aligned along the direction of flow, then θD would be zero and cos θD = 1 and we would
have a colour flow map that showed real velocity of flow along the vessel.
Directional Beam Forming

Using plane-wave beamforming, we can form a beam in any direction, including
parallel to the probe face (Figure 13.19). This technique is known as directional
beamforming [17, 18]. By focusing along a line in the image we create such a beam.
That is, we use the UFUS focused data to look at points along a line in the image.
So, for example, using the stored RF data from the plane wave transmissions, we can
beamform along the line AB by forming focuses along the line AB, as shown. Simi-
larly, we could beamform along line CD, parallel to the probe face. If we beam form
along the line of a blood vessel and use cross-correlation to compare the echo patterns
along that line, we detect the movement along the vessel with no cos θD term, in other
words, we detect vflow along the vessel.
In order to use this procedure, we still need to know the line of the vessel in the
image plane. [19] The way to find this is to create a series of beamformed lines at a
C D
FIGURE 13.19 Illustration of a set of focuses set to form two beams as shown.
A
B
FIGURE 13.20 Using a series of directional beamformed lines to determine the angle of a vessel.
set of angles, e.g. 2° apart, and cross-correlate along each of the lines (Figure 13.20).
The line that yields the highest correlation will be the line that is in line with the
flow. Using the highest correlating values across the image plane yields a flow map in
which the flow velocity and angle of flow is known for all points. By this means, we
can form a vector flow map as shown in Figure 13.21.
The problem with directional beamforming in this way is that it is computa-
tionally very expensive. A whole set of lines need to be beamformed and tested for
maximum correlation each time each image point is refreshed. An alternative tech-
nique allows the angle of flow θD to be determined directly and quickly for each field
point, but gives a slightly lower accuracy for the final value of the flow vector. The
technique is called transverse oscillation and it allows the lateral vx component of
the flow to be determined.
Transverse Oscillation (TO)

As noted above, the normal CDU map shows the value of vz along the axis of the ultra-
sound beam, as cross-correlation between successive pulses shows movement along
the beam axis.
Directional color flow image

25
0.18
30 0.16
0.14
Velocity (m s−1)
35 0.12
Depth (mm)
0.1
40 0.08
0.06
45 0.04
0.02
50
–10 –5 0 5 10
Lateral distance (mm)
FIGURE 13.21 A vector flow map of a vessel at 90° to the surface of the probe using directional
beamforming. Arrows show the vectors of velocity. Source: Jensen and Oddershede [19]/with
permission of IEEE.
The estimation of vx, the transverse velocity component of blood, is a more diffi-
cult problem. One way to find the transverse velocity would be to look for transverse
movement in the speckle pattern across the image from one frame to the next. This
depends on the frame rate and is not particularly accurate.
A better way is to create a ‘pulse’ in which the wavefronts are aligned along the z-axis
and travel out along the X axis, in a similar manner to the way normal pulse wavefronts
are aligned along the X axis and travel down the Z axis. That is, we create a transverse
oscillation of normal longitudinal pressure waves as shown in Figure 13.22.
This then enables the transverse velocity vx to be estimated as changes in the
phase of transverse oscillations, in the same way as conventional CDU estimates vz
from phase changes along the beam axis (see Chapter 9).
The principal of how transverse oscillations are created is as follows (Figure 13.23).[20]
If two beams of ultrasound are transmitted at the same time so they crossover at a point
in the image plane they will interfere with each other to create wavefronts aligned along
the z-axis.
The wavelength of the transverse oscillations in the x-direction is given by
2 T
z
x
d
The x and z components of velocity can then be detected by beamforming –
equivalent to sending out pulses – along three directions. The first beam is sent out
Normal axial Transverse

oscillation oscillation
FIGURE 13.22 Diagram showing the difference between axial and transverse oscillation.
A B A d B
θ z
λT
λx
FIGURE 13.23 The creation of transverse waves by using interference from two crossing beams
A and B. (a) shows how the normal wavefronts perpendicular to each beam interfere to produce
wavefronts in the transverse (x) direction. (b) shows the resultant transverse wavelength λx
compared to the transmitted wavelength λT.
along the z axis and obtains a Doppler sample for vz in the conventional way. The
other two beams are sent out together so as to cross over and create a transverse oscil-
lation. The detected phase – equivalent to the echo back – will give a Doppler sample
of the vx component.
The Doppler signals for fz and fx are estimated by comparing the phase shift between
successive sets of three beamformed pulses using autocorrelation between sample pairs.
Then
2v z fT cos vx
fz and fx
c x
From which the velocity vector for each pixel in the image can be calculated using
vx
D
arctan and v v 2z v 2x
vz
The two interfering beams are apodised to give a Gaussian beam shape (see
Chapter 3) so that side lobes, that could give erroneous information, are reduced and
the beams have a uniform shape at all depths of interest.
Using plane-wave beam forming, the whole field of view can be scanned for z-
axis and x-axis vector information at real-time rates with transverse oscillations being
formed for every field point in the image from the stored RF data.
The resolution of transverse oscillation detected vectors is somewhat less than that
using directional beamformed scan lines. The two can be combined by using TO to
find the vector then using directional beamforming to give the displayed vectors. [21]
Display of Flow Vectors

Determining the magnitude and direction of flow in every part of the image provides
an enormous amount of data and there is the question of how to display this data in a
meaningful way that can easily be appreciated by the sonographer. Two displays that
have been used are a vector display and a colour map display.
Vector Display
Vectors are shown as small arrows superimposed on the blood flow such that
the length and direction of the arrow shows the speed and direction of flow
(Figure 13.24).
Colour Display
Flow direction and speed may be shown using a colour mapping that has direction
shown by colour hue and velocity by colour saturation (Figure 13.25).
Using 2-D transducer arrays, vector flow mapping can be extended to 3-D vector
flow maps (Figure 13.26). Data may be collected across the cardiac cycle enabling flow
velocities to be examined at each stage in the cycle to give 4-D information. [25]
1 m s−1
FIGURE 13.24 A flow vector display using arrows to show flow direction superimposed on a
normal CDU map. Source: Madiena et al. [22]/with permission of IEEE.
0
Red Yellow
+ve
Vz 0 0
–ve
Blue Green
0
–ve +ve
Vx
FIGURE 13.25 (a) Diagram of display scale for flow vectors using colour and saturation to show
direction and magnitude of flow. (b) example of jet flow from an aortic valve AV stenosis into the
ascending aorta AA. Source: Hansen et al. [23]/with permission of Elsevier.
60
60
Vx velocity (cm s–1)
40
30
20
0
4 vx
1
3.5
0.5
3 0 0
vz vy |v|
2.5 –0.5 (cm s–1)
20 cm s–1
Depth (cm) 2 –1 Lateral position (cm)
FIGURE 13.26 Velocity vectors at one point in a time sequence from a 3D scan of an 8mm vessel
at 90° to the probe face. Source: Holbek et al. [24]/with permission of IEEE.
NOTES
• Knowing the velocity vector within the vessel, the scanner can automati-
cally set the correct angle for pulse wave (PW) Doppler waveforms.
• Knowing the velocity distribution across the vessel lumen, the volume flow
(mL min−1) can be calculated.
• Knowing the velocity distribution at all points along a vessel or across a valve
enables pressure gradients within the vessel to be estimated noninvasively.
For example the pressure drop caused by a tight stenosis. With velocity
vector data, this may be performed using the Navier-Stokes equation to give
better estimates of pressure drop than is possible with conventional imaging
(see Chapter 12) [26].
REFERENCES
1. Tanter, M. and Fink, M. (2014). Ultrafast imaging in biomedical ultrasound. IEEE Transactions
on Ultrasonics, Ferroelectrics, and Frequency Control 61: 102–119.
2. The American Registry for Diagnostic Medical Sonography (2022). https://www.ardms.org/dr-
kremkau/ (accessed August 2022).
3. Montaldo, M., Tanter, M., Bercof, F.J. et al. (2009). Coherent plane-wave compounding for very
high frame rate ultrasonography and transient elastography. IEEE Transactions on Ultrasonics,
Ferroelectrics, and Frequency Control 56: 489–506.
4. Bercoff, J. (2011). Ultrafast ultrasound imaging. In: Ultrasound Imaging - Medical Applications
(ed. I.V. Minin and O.V. Minin) [Internet]. London: IntechOpen https://doi.org/10.5772/19729.
5. Jaeger, M., Robinson, E., Akarçay, H.G. et al. (2015). Full correction for spatially distributed
speed-of-sound in echo ultrasound based on measuring aberration delays via transmit beam
steering. Physics in Medicine and Biology 60: 4497–4515.
6. Stähli, P., Kuriakose, M., Frenz, M. et al. (2020). Improved forward model for quantita-
tive pulse-echo speed-of-sound imaging. Ultrasonics 108: 106168. https://doi.org/10.1016/j.
ultras.2020.106168.
7. Rau, R., Schweizer, D., Vishnevskiy, V. et al. (2019). Ultrasound aberration correction based on
local speed-of-sound map estimation. IEEE International Ultrasonics Symposium 2019: 2003–2006.
https://doi.org/10.1109/ULTSYM.2019.8926297.
8. Jaeger, M., Held, G., Peeters, S. et al. (2015). Computed ultrasound tomography in echo mode
for imaging speed of sound using pulse-echo sonography: proof of principle. Ultrasound in
Medicine and Biology 41: 235–250. https://doi.org/10.1016/j.ultrasmedbio.2014.05.019.
9. Bercoff, J., Montaldo, G., Loupas, T. et al. (2011). Ultrafast compound doppler imaging: pro-
viding full blood flow characterization. IEEE Transactions on Ultrasonics, Ferroelectrics, and
Frequency Control 58: 134–147.
10. Jensen, J.A., Nikolov, S.I., Hansen, K.L. et al. (2019). History and latest advances in flow estima-
tion technology: from 1-D in 2-D to 3-D in 4-D. IEEE International Ultrasonics Symposium 2019:
1041–1050. https://doi.org/10.1109/ULTSYM.2019.8926210.
11. Demené, D., Deffieux, T., Pernot, M. et al. (2015). Spatiotemporal clutter filtering of ultrafast
ultrasound data highly increases doppler and fUltrasound sensitivity. IEEE Transactions on
Medical Imaging 34: 2271–2285.
References 299
12. Zhang, X., Zhang, L., Li, N. et al. (2019). Vascular index measured by smart 3-D superb microvas-
cular imaging can help to differentiate malignant and benign breast lesion. Cancer Management
and Research 11: 5481–5487.
13. Bercoff. J., Chamak-Bercoff. J., Fraschini. C. et al. (2011). UltraFast™ doppler. Supersonic
Imaging White Paper.
14. Malonea, A., Cournane, S., Fagan, A. et al. (2019). Investigation of intrinsic spectral broadening
on the Aixplorer ultrafast ultrasound system. Physica Medica 67: 210.
15. Jensen, J.A., Nikolov, S.I., Yu, A.C.H. et al. (2016). Ultrasound vector flow imaging: part I:
sequential systems. IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control 63:
1704–1721. https://doi.org/10.1109/TUFFC.2016.2600763.
16. Jensen, J.A., Nikolov, S.I., Yu, A.C.H. et al. (2016). Ultrasound vector flow imaging: part II:
parallel systems. IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control 63:
1722–1732. https://doi.org/10.1109/TUFFC.2016.2598180.
17. Jensen, J.A. and Nikolov, S. (2002). Transverse flow imaging using synthetic aperture direc-
tional beamforming. IEEE Ultrasonics Symposium Proceedings 2002: 1488–1492. https://doi.
org/10.1109/ULTSYM.2002.1192586.
18. Jensen, J.A. and Nikolov, S.I. (2004). Directional synthetic aperture flow imaging. IEEE
Transactions on Ultrasonics, Ferroelectrics, and Frequency Control 51: 1107–1118.
19. Jensen, J.A. and Oddershede, N. (2006). Estimation of velocity vectors in synthetic aper-
ture ultrasound imaging. IEEE Transactions on Medical Imaging 25: 1637–1644. https://doi.
org/10.1109/TMI.2006.883087.
20. Jensen, J.A. and Munk, P. (1988). A new method for estimation of velocity vectors. IEEE
Transactions on Ultrasonics, Ferroelectrics, and Frequency Control 45: 837–851.
21. Jensen, J., Hoyos, V., Stuart, C.A. et al. (2017). Fast plane wave 2-D vector flow imaging using trans-
verse oscillation and directional beamforming. IEEE Transactions on Ultrasonics, Ferroelectrics,
and Frequency Control 64: 1050–1062. https://doi.org/10.1109/TUFFC.2017.2693403.
22. Madiena, C., Faurie, J., Poree, J. et al. (2018). Color and vector flow imaging in parallel ultra-
sound with sub- nyquist sampling. IEEE Transactions on Ultrasonics, Ferroelectrics, and
Frequency Control 65: 795–802. https://doi.org/10.1109/TUFFC.2018.2817885.
23. Hansen, K.L., Moller-Sorensen, H., Kjaergaard, J. et al. (2016). Intra-operative vector flow
imaging using ultrasound of the ascending aorta among 40 patients with normal, stenotic and
replaced aortic valves. Ultrasound in Medicine and Biology 42: 2414–2422.
24. Holbek, S., Christiansen, T.L., Stuart, M.B. et al. (2016). 3-d vector flow estimation with row–
column-addressed arrays. IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency
Control 63: 1799–1814. https://doi.org/10.1109/TUFFC.2016.2582536.
25. Correia, M., Provost, J., Tanter, M. et al. (2016). 4D ultrafast ultrasound flow imaging: in vivo
quantification of arterial volumetric flow rate in a single heartbeatPhys. Medical Biology
61: L48–L61.
26. Olesen, J.B., Villagómez, H.C.A., Møller, N.D. et al. (2018). Non-invasive estimation of pressure
changes using 2-D vector velocity ultrasound: an experimental study with in-vivo examples.
IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control 65: 709–719. https://doi.
org/10.1109/TUFFC.2018.2808328.
CHAPTER 14
Elastography
The tissues in the body that are loosely termed ‘soft tissues’ vary in their stiffness/
rigidity or their elasticity. And the stiffness of the tissue may change when it becomes
diseased, for example, a malignant tumour is often stiffer than surrounding tissue,
or in response to physiological changes such as oedema. Measuring tissue stiffness
or elasticity therefore provides another ‘window’ into the body to form a diagnosis.
Elasticity is a measure of how much the tissue moves in response to an applied
force. Rigidity or stiffness is a measure of how much the tissue resists moving in
response to an applied force,
1
i.e. elasticity
rigidity
The basic principle behind elastography is that we apply a deforming force to cre-
ate a distortion in the tissue and from the way it responds we can infer its mechanical
properties and display the results, for example, as a colour mapping on a B-mode image.
For an everyday example, to see how this principle works, imagine pressing down
on the surface of a trifle in a glass bowl with a spoon. The custard on the top has very
little rigidity and moves easily. The jelly requires more force to squeeze it and the cake
in the jelly is relatively rigid and does not distort much at all.
Elastography with ultrasound can be thought of as a more sophisticated extension
of manual palpation, which is regularly used by physicians to examine patients.

301
302 Chapter 14 Elastography
BACKGROUND THEORY
Within a time frame of <1 second, soft tissue may be considered incompressible. If a
force is applied in one direction, the dimension of the tissue on that axis will decrease
but the volume remains the same as there will be a corresponding spreading out in a
plane lateral to the applied force, known as the Poisson effect (Figure 14.1). When
the force is removed, the tissue rebounds to its original shape. If the force is sustained,
then the property of incompressibility may no longer be true. For example, when com-
pressing oedematous tissue, fluid may be driven out of the compressed tissue into the
surrounding tissue space and the oedematous tissue then has a reduced volume until it
refills and induration of the skin may be noticed.
Assuming soft tissue is incompressible, if we press down on the skin, the tissues
will move a certain amount for a particular applied force. The deformation will vary
from one tissue to another depending on how stiff it is. The distance along the axis of
the applied force will decrease and the tissue will spread out and expand in a direction
perpendicular to the applied force as described.
The applied force is called the stress σ and the resulting change in length along
the axis of the applied force, l is called the strain (Figure 14.2). The ratio of the two
l
is called Young’s modulus E.
stress applied force
E
strain resulting movement
A
B
FIGURE 14.1 The motion of tissue from A to B under an applied unidirectional force assuming
it is incompressible.
F=σ Δl
l
FIGURE 14.2 Definition of Youngs modulus: change in length Δl in response to applied force σ.
E has a high value for a stiff rigid material (small strain) and a low value for a soft
deformable material (large strain).
So, by applying a force and using the change in position of the target tissue, measured
from the returning ultrasound echoes, we can estimate the stiffness of tissues in the body.
ELASTOGRAPHY
The general term elastography is used to cover a set of closely related ultrasound tech-
niques that give information on tissue stiffness. These fall into two groups: those that
estimate the strain resulting from an applied force are called strain elastography
(SE), and those that create a sideways moving displacement or shear wave inside
the tissue, and measure the velocity of the shear wave to estimate tissue stiffness, are
called shear wave Elastography (SWE). [1, 2]
METHODS OF APPLYING THE DISTORTING FORCE
In order to measure tissue stiffness with ultrasound, we need to apply a force to distort
the tissue. There are three main ways to achieve this:
1. Push or vibrate the surface of the body with the ultrasound probe or tool such
as a probe footplate extender.
2. Use distortions created by normal physiological movements, e.g. pulse, respi-
ration, or heartbeat.
3. Use acoustic radiation force from an ultrasound pulse to remotely create a
‘push’ inside the tissue.
The results obtained may be qualitative – showing relative stiffness, or they may
be quantitative – giving actual stiffness as a numerical value.
STRAIN ELASTOGRAPHY (SE)
Uses the change in position from an applied force to estimate strain.

The applied force is provided either by manual compression on the skin, e.g. by
pushing down on the transducer probe, or by tissue displacement from the heart or
respiration, as may be used when looking at breast lesions.
SE with Manual or Physiological Palpation

If a constantly increasing force is applied such that the strain increases uniformly,
speckle tracking across successive frames can detect the movement of the tissue.
Speckle Tracking
The speckle pattern, produced by the interference of microscopic scatterers within

the length of the ultrasound pulse, will remain static as long as the tissue and
probe remain stationary. If the tissue moves, even by a fraction of a wavelength,
the speckle pattern will change as a new interference p attern emerges from a
new pattern of scatterers. However, as long as the movement in tissue is small, of
the order 1 wavelength, the speckle pattern from a particular group of scatterers
will be virtually unaltered but will occur at a slightly different time in the pulse-
echo sequence (Figure 14.3). Speckle tracking compares the received RF signal
from consecutive pulses to find the speckle pattern from the first pulse in the
second and so determine how far the speckle pattern has moved using
1540. t
d
2
where Δt is the time between pulses.

Tracking the speckle in this way can be achieved either by using autocorrelation to
find the best RF pattern match between pulses, or with Doppler techniques which
look for the phase difference between consecutive pulses.
Speckle tracking may be carried out in the axial direction by comparing pulses
along the same beam or in the lateral direction by comparing pulses from adja-
cent beams.
P1
ΔTime
1
=
P2 PRF
P3
Δz
Velocity =
Δz Δz Δt
FIGURE 14.3 Speckle tracking across three consecutive pulses P1–P3.The change in position
is detected by looking for identical patterns in the RF signal, from which the distance or
velocity of movement can be estimated.
F1 F2
Depth z
t1
t2 Δz
Reference Search
window window
FIGURE 14.4 Speckle tracking to determine strain. The RF signal from a reference window in
the first frame F1 is looked for within a search window in the second frame F2.
Using the RF echo signal and cross-correlation, a reference window from the first
frame is compared across the search window in the second frame to find the best fit
(Figure 14.4). For each scan line, the axial distance Δz that the reference window
signal has moved is the displacement caused by the applied force. The procedure is
performed for all depths along each scan line so as to cover the region of interest. This
is repeated using successive pairs of frames with frame averaging, to improve the esti-
mated value of the displacement (Figure 14.5). (Note: this is not the same window as
the elastogram box seen on the B-mode image)
From the displacement obtained from the strain-estimating window or
tracking window, the strain is then calculated as the gradient of displacement, as is
seen in Figure 14.6.
d
strain . 100% i.e. the gradient as a percentage change, where d0 is the original
d0
spacing of the echo pattern.
The size of the window is a compromise between resolution and signal to noise
ratio (SNR), and is a user control on some machines.
Strain
Frame 1 Frame 2
image
Δ movement
FIGURE 14.5 The principle of comparing two successive images to look at movement in the
image frame.
d0 S
d0 M Δd5
d0 H Δd4
d1
d2 Δd3 Δd S S
d0 S Δd2 M d
d3 H S M M
M d4 Δd1 M
d0 S H
d5
d0 d0 d0 d0 d0 d0 d0 d0 d0 d0
(a) (b) (c)
FIGURE 14.6 (a) The change in thickness of a tissue with 5 layers, thickness d0, but whose
stiffness varies (H hard, M medium and S soft) when a stress σ is applied. (b) The change in
thickness (Δd1 = d0−d1) graphically. (c) The gradient for each layer shows the strain.
On compression by the transducer, there may be tissue displacement that is lat-

eral to the beam axis but still within the scan plane. That is, axial compression causes
some lateral expansion that causes displacement noise (see Figure 14.1). The scanner
therefore performs cross-correlation in the lateral and axial directions (i.e. across
beams Δx as well as along beams Δz) to improve the precision of the axial displace-
ment estimate, but only retains the axial component for the elastogram. This reduces
noise arising from lateral tissue movement.
The resulting strain data may be displayed as a colour map or a greyscale map
displayed in an elastogram box on the B-mode image. The map is known as an
elastogram. The size of the elastogram box may be chosen by the user. In order to
improve the quality of the elastogram, frame averaging is applied.
The use of a footplate extender on the transducer can improve the uniformity of
the applied stress and maximise the depth to which it penetrates (Figure 14.7)
FIGURE 14.7 Probe footplate extender. Source: Dietrich et al. [3]/Thieme Medical
Publishers, Inc.
User Controls 307
NOTE
There is not a standardised colour scheme and colours may vary between scan-
ners, e.g. red-green-blue – may be hard/soft or soft/hard. USER BEWARE!.
USER CONTROLS
NOTE
Not all scanners will have these controls accessible to the user.
Frequency
The ultrasound frequency used for the elastogram may be chosen independently of
the B-mode frequency. As for B-mode, a lower frequency will give greater penetration
for sensing the strain but poorer resolution. Reducing the frequency also reduces the
elastogram frame rate.
Frame Rate
Some scanners have a separate control that adjusts the elastogram frame rate to enable
the user to match the compression speed and amplitude.
Tracking Window
The tracking window used for speckle tracking may be adjusted. A larger window
gives a better signal to noise ratio but reduces the resolution.
Frame Reject
Lower quality elastogram frames are rejected and colour is not shown in the elasto-
gram box for these frames. This can lead to a ‘flashing’ effect on the elastogram as
frames are rejected or accepted. Lowering the reject threshold reduces flashing but
also reduces the quality of the elastogram.
Noise Reject
A filter that removes noisy pixcels within each frame by rejecting regions where the
echo signal is not strong enough for speckle tracking correlation, for example, in cysts
and other hypoechoic areas. Colour is not displayed in rejected pixels.
Line Density
A lower line density results in a higher frame rate but with lower spatial resolution.
Frame Averaging
Allows fames to be blended with previous frames to improve image quality. It
reduces flashing.
Dynamic Range
Adjusts the levels at which significant colour changes indicating strain levels are
made. With higher values of dynamic range only more extreme values of strain are
shown in red or blue with more of the image showing green.
Colour Map
The colour map used to display the elastogram may be changed.
Colour Blend
Alters the transparency of the elastogram box over the B-mode image.
NOTES
• Hand compression motion must be in line with the axis of the ultrasound
beams. If there is movement along a line that runs out of the scan plane
or twists with respect to a fixed scan plane then cross-correlation between
frames will be lost and the elastogram will be noisy or dropout.
• Uneven stress across the footprint of the transducer leads to non-uniformity
in the elastogram. Therefore ‘heel and toe’ movements of the transducer
should be avoided.
• Curved array transducers present a problem due to tissue under the centre of
the transducer experiencing higher stress than that at the ends of the trans-
ducer. This can be mitigated by the use of a probe footplate to increase the sur-
face area of the transducer and give more consistent pressure along its length.
• The best quality data comes from echoes where there is near constant rate
of surface displacement, not at the extremes of displacement, as shown in
Figure 14.8.
• The strain images should be optimised by the sonographer varying the com-
pression speed (0.5/second–2.0/second), the amplitude (1–3 mm) of the
compression and the persistence or frame averaging used. These should
User Controls 309
Best data
Time
FIGURE 14.8 The best SE data comes from uniform compression of tissue.
be adjusted interactively whilst viewing the image so as to best display the

tissue strain contrast between hard and soft tissue. The image frame rate
may also be altered to achieve best strain contrast.
• Some scanners have a quality indicator giving a numerical or visual display
of the amount of compression-release being applied to aid the operator in
finding the optimal technique. More recently indicators of pressure gradient
across the transducer footprint have become available enabling the operator
to check they are applying an even pressure.
• Aim for consistency in strain seen across frames.
• The size of the elastogram box and the line density may be adjusted to
improve frame rate. However, the box should include enough normal tissue
so that relative changes to normal, in abnormal tissue, can be observed. The
suspected lesion should take up <50% of the elastogram box [3].
• Care should be taken when adjusting the frame reject and noise reject controls
as they determine the information that is, or is not, seen in the strain image.
• The strain image shows relative differences in stiffness and the information
is qualitative. Comparison should not be made across different manufac-
turers scanners.
• Strain imaging works best where there are focal changes in tissue stiffness
within the field of view. It shows relative contrast between tissue stiffness
and therefore gives little information where there is a generalised change in
tissue stiffness such as is seen in cirrhotic liver, where the lack comparator
tissues means relative strains are not seen.
• Imaging at depth depends on the transmission of the palpation into the
tissue. The displacement reduces with depth but the strain, or percentage
deformation, will be constant with depth. A deeper scan may necessitate a
slower rate of probe compression and hence reduce frame rate.
The small tissue movements resulting from cardiac pulsation and respiration may
also be tracked to produce strain images, e.g. in breast tissue.
SE ARTEFACTS
Care needs to be taken in interpreting strain images as a number of physical features

can affect the elastogram causing artefacts.
• Tissue is non-linear in the relationship between the force applied and the
deformation produced leading to artefacts such as stress concentration artefact.
• Elasticity in tissue is often anisotropic and probe orientation with respect to
the tissue may alter the result.
• The tissues in the body are mechanically discontinuous when interrupted by
anatomical features such as fluid collections, organ or tumour boundaries and
scars. These may produce artefacts on the strain image.
• As tissue is compressed it becomes more rigid and pre-compression or pre-
loading by applying too much pressure with the probe prior to starting com-
pressions, will diminish the strain contrast and produce a noisy image. The
pressure should be only just sufficient to obtain a diagnostic B-mode image.
A generous layer of lubricating gel will enable good contact with little initial
pressure and good B-mode to be obtained.
The graph in Figure 14.9 shows the effect of pre-compression on strain con-
trast in breast imaging. Cancerous tumours are stiffer than fat. If the breast tissue
A B C D
12
10
C
8
Vs (ms–1)
4
F
2
0
0 10 15 20 25 27.5 30 35 40 50
Total % compression from body surface
FIGURE 14.9 The effect of pre-compression on shear wave velocity vs in the breast for fat
(F) and cancer (C). In zone A (<10% pre-compression) benign and malignant tissue may be
distinguished on strain and shear wave images. In zones B-D the signal degrades. Strain images
become too noisy to use and shear wave velocities become indistinguishable. Source: Adapted from
Barr and Zhang [4].
SE Artefacts 311
is compressed by the probe so that the measured thickness of the breast tissue from
skin to a fixed point (e.g. rib) is reduced by 40%, the fat shows the same stiffness as the
tumour. The effect on the elastogram is shown in Figure 14.10.
As a result of these factors, it is recommended that the following be observed
where possible when performing SE.
• Strain imaging is only going to be reliable if there is a good quality B-mode

image. Therefore, ensure you have a good B-mode image.
• Targets closer to the transducer are imaged more reliably.
• Targets surrounded by near homogenous tissue are imaged more reliably than
where there are closely adjacent tissue boundaries.
• There should be no anatomical planes that allow slip movement anterior to or
within the region of interest (Figure 14.11).
(a)
(b) (c)
FIGURE 14.10 SW Elastogram showing the effect of pre-compression. (a) B-mode of a benign
fibroadenoma, (b) its shear wave velocity when done properly is 2.4 ms−1 and (c) with heavy
pre-compression rises to 5.1 ms−1. Note also, the shear wave velocity of surrounding fat also
increases with increased compression. Source: (b) and (c) Barr [5]/Korean Society of Ultrasound
in Medicine.
Slip plane
FIGURE 14.11 Diagram of a tissue boundary that may form a slip plane when performing SE.
• Check the B-mode image to make sure there is no lateral movement of the
tissue with the compression or breathing, etc.
• There should be no structures nearby that would damp the shear stress, e.g.
large veins.
• Aim for a limited number of targets in the elastogram field of view so it covers
normal and abnormal tissue (target <50% of elastogram box area). [6]
The following artefacts may be seen on SE. Some of these artefacts can be
diagnostically useful or alert the sonographer to technical problems in the scan.
Stress Concentration Artefacts

Soft tissue strain is greater when it is adjacent to hard tissue and the inclusion of
a hard tissue in a soft background may cause the surrounding soft tissue to appear
softer than it is where stress concentration occurs as shown in Figure 14.12.
Edge Enhancement Effects

Slippery boundaries between structures may cause edge enhancement effects to be
seen on the elastogram. This occurs where, upon palpation, tissue on one side of a
boundary slips relative to the tissue on the other side (Figure 14.11). Or, it may occur
when a solid structure is not strongly tethered and moves with respect to surround-
ing tissue with the compressions (Figure 14.13). As a result, the edge may appear
enhanced and there may be slight displacement of strain boundaries relative to the
corresponding structures on the B-mode image.
Related to these two artefacts, care needs to be taken to ensure that the image
is not being affected by the stiffness of structures immediately outside the imaging
plane, e.g. boundaries, cysts, low/high strain structures.
SE Artefacts 313
FIGURE 14.12 Example of stress concentration artefact. Elastogram of a hypoechoic thyroid

nodule shows the tissue distal to the stiff (red) nodule as softer (blue) than the surrounding tissue.
Source: Courtesy of Stephen Klarich.
FIGURE 14.13 A greyscale elastogram of a superficial benign breast lesion slipping within
surrounding fatty tissue as the probe is palpated giving rise to an artefactual bright ring of
apparently soft tissue. Source: Barr [5]/Korean Society of Ultrasound in Medicine.
Appearance of Cysts
A watery fluid is incomprehensible (try compressing a syringe full of water).
A fluid filled cyst may distort when compressed depending on the stiffness of
the tissue encapsulating it. However, fluid-filled cysts do not return echoes and
appear black on B-mode. In strain imaging the lack of echoes and speckle to track
produces noise and artefacts. In particular they produce what is known as blue/
green/red artefact.
Push Artefact
High strain
Soft
z Δz
Cyst Low strain
z
Soft
Depth Depth
Cyst
(a) (b) (c) Cyst
FIGURE 14.14 Illustration showing how a cyst produces ‘blue/green/red’ artefact in SE. (a) the
tissues, (b) movement detected along the ultrasound beam, (c) the gradient of movement showing
the resulting artefact.
Blue/Green/Red Artefact
Figure 14.14 shows how the gradient of strain changes at the boundaries of a non-echogenic
cyst to produce blue/green/red artefact. It gets its name from the three-colour layering
pattern seen in cystic lesions. Called blue/green/red artefact (blue = stiff), the
three colours form three bands across the whole cyst as shown in Figure 14.15.
Shadowing
As with other ultrasound techniques, acoustic shadowing will cause loss of information
and a noisy image in shadowed areas.
ACOUSTIC RADIATION FORCE IMPULSE IMAGING (ARFI IMAGING)
ARFI uses the third way of creating a distortion in tissue, namely to use ultrasound
pulses to create a disturbance directly inside the tissue mass. When an ultrasound
pulse travels through tissue, energy is absorbed from the pulse and the tissue expe-
riences a force known as radiation force (see Chapter 11). For the ARFI technique
a higher intensity ‘push pulse’ is transmitted into the tissue, with a focused beam
centred on a region of interest. This generates a disturbance in the tissue created by
the radiation force from the push pulse. It produces a displacement of a few microns
in the axial direction which recovers after a few milliseconds. As before, the amount
of tissue movement this radiation force produces, is detected using speckle tracking or
Doppler detection. By transmitting a series of these push pulses, followed by normal
pulses along the same line to interrogate the tissue movement, and stepwise across
Acoustic Radiation Force Impulse Imaging (ARFI Imaging) 315
FIGURE 14.15 Illustration of ‘blue/green/red’ artefact in the cyst indicated by the arrow,
only in this case it is red/green/blue as red = stiff in this elastogram. Source: Courtesy of
Stephen Klarich.
the field of view, a region of interest can be scanned and the strain information be
displayed using a colour mapping. A typical push pulse will be a long burst (10’s of μs)
of focused ultrasound.
Pros and Cons of ARFI

ARFI has a number of advantages over manual palpation.
Pros:
• Push pulses can be delivered deep into tissue.
• Delivers more uni-axial pulses which are less user dependent.
• Gives better resolution elastograms.
• Less strain-concentration artefacts.
• Cysts do not have stress cone bright ups anterior and posterior to the cyst.
• Better strain contrast and improved SNR at depth.
• Less influenced by slip movements anterior to the target of interest.
Cons:
• May be affected by the absorption and reflection of push pulses.
• Important not to push on the skin surface and pre-compress the tissue.
• The depth of imaging is limited by the maximum size of push pulse that can be
safely transmitted into the body.
STRAIN RATIO
Strain imaging itself is qualitative, showing relative differences in stiffness. A semi-

quantitative approach can be achieved by comparing the strain shown in a target
tissue with that of a tissue whose strain is used as a normal reference value. This then
forms a strain ratio. Most malignant tissue is stiffer than normal tissue and the strain
in a suspected tumour may be compared with the strain in an adjacent area of fat
within the same field of view, for example.
NOTES
• Ratios vary between scanners and values should not be compared across
different manufacturers machines.
• The reference region of interest should be positioned to the side of the target
region of interest, rather than above or below it, so the pushing conditions
are similar at that depth.
• Measured ratios have greater errors than theoretical values for softer inclu-
sions. Therefore, be very cautious when using ratios. In practice they can
only reliably used to make a binary analysis, i.e. >1 harder or <1 softer.
SHEAR WAVE ELASTOGRAPHY (SWE)
Background Theory
We saw in Chapter 1 that normal ultrasound waves are longitudinal pressure waves
that travel through soft tissue at the speed of sound 1540 ms−1. The tissue will move
in response to these pressure waves, but we cannot use them to directly measure elas-
ticity because they travel too fast, traveling at the same rate as the ultrasound pulses
we might use to interrogate them.
For quantitative measurement of tissue elasticity with ultrasound, the solution is
to make use of shear waves. Consider one of those ‘slinky’ spring toys (Figure 14.16).
If we stretch it out and move the end backwards and forwards in the direction of the
line of the spring, we see a longitudinal wave travelling along the spring, exactly like
a normal sound wave. We can induce another sort of wave in the spring by moving the
end sideways relative to the axis of the spring, similar to shaking a rope from side to
side. This is called a transverse wave. When such a transverse wave occurs within a
solid it is called a shear wave.
NOTE
Here, transverse or shear waves are a different phenomenon from transverse oscil-
lations used for vector flow mapping (Chapter 12). Those used for vector flow map-
ping are longitudinal compression waves travelling out along the x-axis.
Longitudinal wave
Transverse
or shear wave
FIGURE 14.16 Longitudinal wave and transverse wave seen by oscillating a ‘slinky’. In both
cases the wave travels along the slinky from one end to the other.
In a solid material, such transverse oscillations of molecules push and pull on

those molecules next to them and the shear wave will travel out through the material
from the source of the wave (Figure 14.17). The particle displacement is perpendic-
ular to the direction of propagation and the speed of these shear waves is about 1000
times slower than the 1540 ms−1 speed of sound of longitudinal waves, i.e. 1–10 ms−1.
To picture what goes on in tissue and how we might detect shear waves think of a
stone dropped into a still pond. The waves on the surface of water are like shear waves.
Their oscillation is perpendicular to their direction of travel. The stone is the source
of the waves and the waves travel out from the source with the surface going up and
down. A cork sitting on the surface some distance away from the source will start to
bob up and down with the expanding wave as the wave reaches it and passes it. The
motion of the cork is in the same vertical direction as the stone travelled that started
the waves. So, for our ultrasound case, the molecules will oscillate in the z-direction
but the wave is travelling out in the x-direction. We can therefore use normal ultra-
sound pulses to detect this motion in the z-direction using speckle tracking, and find
the time it takes for the shear wave to reach a given point away from the source and so
find the shear wave velocity in the x-direction (Figure 14.18).
Within the soft tissues found in the body, shear waves do occur, but are very much
weaker than the normal longitudinal waves as soft tissue cannot support transverse
waves very well. However, shear waves are more sensitive to tissue stiffness than
longitudinal waves and they can provide the ‘probe’ to investigate tissue stiffness. In
particular the speed of sound of the shear wave can be used to differentiate the
stiffness of soft tissues.
Recall from Chapter 1, the speed of sound of the normal longitudinal wave is given by
K
c
where K is the bulk modulus and ρ the density of the material.

vs
x
FIGURE 14.17 Diagram of a shear wave in a solid showing the transverse motion of molecules
whilst the wave propagates in the direction show by the arrow.
Push B-mode
pulse pulse
Shear vs
wave
x
Longitudinal
wave
(a) z (b) Δx
FIGURE 14.18 Diagram showing (a) the shear waves generated by longitudinal push pulse
oscillating in the z direction whilst propagating outward in the x direction and (b) a B-mode pulse
to interrogate the shear wave at time Δt for distance Δx to find shear wave velocity vs = Δx/Δt.
V
The bulk modulus is the ratio of the change in change in volume to a change
in applied pressure ΔP. V
P
K V see Figure 1.20
V
In a similar manner, the speed of sound for shear waves cs is given by
G
cs
where G is the shear modulus, the ratio of the applied force to the resulting distor-
tion (Figure 14.19).
transverse force
G
resulting distortion
Δz
FIGURE 14.19 Definition of shear modulus showing distortion Δz caused by an applied force F.
Shear wave speeds of sound cs in soft tissue range from 1–10 ms−1 and G has a
large range.
Young’s modulus E and shear modulus G are related
E DG D cs2
where D is a constant of proportionality.

For the soft abdominal tissues D = 3, i.e. for these tissues
E 3G 3 cs2
So, if we measure the shear wave speed of sound cs, we can estimate E and G in
kPa (or Nm−2) and obtain quantitative estimates of tissue stiffness.
For SWE the applied force used is acoustic radiation force ‘push pulses’, as used
in ARFI SE. However, instead of performing speckle tracking of movement along the
same beam axis that the push pulse was sent out, speckle tracking is performed along
the beam axis (Δz) of B-mode beams sent out in pulses moving laterally away from
the push pulse axis, as shown in Figure 14.20. As tissue is disturbed by the push pulse,
shear waves are generated, disturbing the tissue laterally, away, from the axis of the
D2 D2
D1 D1
P
Detecting
Push pulse pulses
FIGURE 14.20 Illustrating the pulsing regime used for point SWE with a push pulse followed by
detecting pulses to monitor the shear wave travelling away from the push pulse.
push pulse beam, with the disturbance moving out at the shear wave speed of sound.
The wavelength of the shear waves is about 1 cm and depends on the width of the tissue
directly excited by the push pulse [7]. The resolution of the elastogram depends on the
resolution of the interrogating B-mode beams and is about 1 mm. The speed at which
the shear waves move can be tracked across a band 5–10 mm wide using the B-mode
pulses to detect the transverse oscillation of tissue (Δz), as the shear wave passes by.
POINT SWE (PSWE)
This technique can give a shear wave speed of sound measurement for a single uni-
form region of interest with a single push pulse beam and detection of resulting shear
wave motion, hence point SWE. Each measurement is static and not updated. It may
be used to give a quantitative single value measure of liver, or other homogenous
organ, stiffness (Figure 14.21).
As the actual shear wave speed of sound is measured, the information may be
given as the shear wave speed of sound or, using the equation
E 3 cs2 for soft tissue,
Young’s modulus can be calculated and the information given as the stiffness
(E) in kPa.
An extension of this technique is to repeat the push-pulse – detection-pulse
sequence for a range of depths and across a region of interest (ROI) to give a static
shear wave elasticity image as a colour elastogram box on the B-mode image (Figures 14.22
and 14.23). The image is static in that it is a single image that is not updated. Using
this technique, the speed of shear waves can be measured to a depth of ~8 cm.
FIGURE 14.21 Point SWE in a patient with splenomegaly. The sample volume position is
indicated by the cursor, the average value of shear wave speed in the sample is shown on the right.
Source: Pereira et al. [8]/with permission of Elsevier.
Point Swe (PSWE) 321
D1 D1 D2 D2
D1 D1 D2 D2
P1 P2
Detecting
pulses
Push pulse
FIGURE 14.22 Typical pulsing regime across an array used to form a SWE image.
FIGURE 14.23 Shear wave elastogram of liver showing the elastogram on the right and a quality
indicator on the left. Average value of stiffness within the circle shown bottom left.
A number of push pulses are transmitted, focusing at different depths within the
elastogram box and across its width. As these deliver a significant amount of energy
to the tissue there needs to be a rest period between each refresh sequence. This is
in order to keep the tissue exposure within safe limits and to prevent the transducer
from overheating (see Chapter 11). The frame rate is therefore determined by the
acquisition time and the cooling time. Using multiple simultaneous transmit pulses
across the ROI and interpolation of acquired values, the elastogram can be refreshed
at a faster rate. Typically, the acquisition time may be 100 ms and the cooling time 2–3
seconds, meaning the cooling time is the dominant factor in determining frame rate.
As with SE, a quality indication may be displayed to allow the user to gauge the reli-
ability of the elastogram.
NOTES
• For other tissues, e.g. kidney and muscle D does not equal three and in
these cases the shear modulus G can be estimated and used instead of
Young’s modulus.
• The shear modulus G also has units kPa.
• For all SWE techniques the results are less accurate for depths below 6cm
due to attenuation of the push pulse generating the shear waves.
SUPERSONIC SHEAR IMAGING (SSI)
This is the most advanced form of shear wave imaging and gives a real-time 2-D or
3-D map of tissue stiffness refreshed several times a second.
As for conventional SWE, SSI uses a series of ARFI push pulses to generate shear
waves. A set of push pulses are sent out along a single transmit beam focused at
increasing depths along that line (Figure 14.24). The rate of transmission of these
push pulses is such that the foci, generating shear waves, move along the transmit line
faster than the speed of the shear waves themselves. Hence the name – supersonic
shear imaging. This creates a column of shear waves moving out from along the
whole length of the push pulse beam in the shape of a cone called a Mach cone, in
the manner of a wake seen behind a moving boat or a sonic boom moving out behind a
supersonic plane. The shear waves produced are then interrogated by speckle tracking
pulses using the ultrafast ultrasound technique, for example, plane wave imaging
with post transmit-receive beam forming (Chapter 13).
PP1
6 ms–1
PP2 Plane shear
waves 2 ms–1
PP3
FIGURE 14.24 Using a series of push pulses PP1–3 generated at supersonic speed to produce a
Mach cone of shear waves.
Shear Wave Compounding 323
Reference image
Time
0 PP1 PP2 PP3 30 ms
Ultra fast imaging

sequence
FIGURE 14.25 A typical push pulse (PP) sequence used for SSI, showing the B-mode reference
images, push pulse sequences and imaging sequences detecting the shear wave velocities.
The following pulsing sequence is typical (Figure 14.25). A reference B-mode

image is created to give the static tissue state against which the shear wave movement
is compared. A series of push-pulses are sent out to create a Mach cone of shear waves
moving out from the beam axis. The shear waves are then tracked using speckle
tracking from beams adjacent to the push-pulse beam axis using ultrafast plane-
wave imaging. This sequence is repeated for a series of push-pulse lines across the
image plane to cover the region of interest. The speed of sound data for shear waves is
used to calculate stiffness and elastograms are produced at the rate of 3-4 Hz. Frame
averaging is used to reduce noise (Figure 14.26).
SHEAR WAVE COMPOUNDING
The ratio of the speed of travel to the speed of sound is called the Mach Number, for
example, a plane travelling at twice the speed of sound is travelling at Mach 2. The
angle at which the shear wavefront moves away from the transmit line depends on
the Mach number of the transmit push pulses along the transmit line. By generating
transmit lines with different Mach numbers, shear waves are generated at different
angles and the elastograms from these transmissions can be averaged to give shear
wave compounding (Figure 14.27). This is similar to B-mode spatial compounding
(see Figure 7.6).
NOTES
• Shear wave elastography is good for assessing diffuse disease as it gives
quantitative values from areas in which the change in stiffness from one
point to another is small. In other words, there are no normal adjacent
regions for comparison.
• As it is quantitative, it allows changes over time to be evaluated.
• SSI is insensitive to patient motion due to the fast acquisition time
• There is a limitation on the depth of field of view (typically 6 to 8 cm) as it

is difficult to safely generate shear waves in deep tissue and the signal gets
noisy. The push pulse intensity would have to be too high to go deeper.
• Caution should be applied when comparing results across different manu-
facturers scanners. [10, 11]
• For shear wave elastography the probe should be held completely still with
little pressure on the skin (see Figure 14.10).
• Fluids do not support shear waves. Cysts and other focal lesions tend to
produce artefacts on shear wave images, as described above. For this reason,
vessels should be avoided within the image.
(a)
t = 1 ms t = 4 ms t = 5 ms t = 6 ms
(b) (c) (d)
–60 –30 0 0 12 24 Soft Hard

dB E (kPa)
FIGURE 14.26 Images from a phantom with a 10 mm stiff inclusion showing (a) the Mach cone
arising from a set of supersonic push pulses at increasing time intervals after generation. The
stiff inclusion is seen to distort the shear wavefront. (b) a normal B-mode image of the phantom
where the inclusion is hardly visible. (c) The shear wave elastogram of the inclusion and (d) a
strain elastogram of the same phantom using manual palpation of the probe. Source: Fink and
Tanter [9]/with permission of AIP Publishing LLC.
The estimate of the shear wave velocity vs depends on reliable acquisition (absent
patient/probe movement) and reliable data processing (low SNR). Based on detecting
movement and noise from the ultrasound data, a quality measure may be estimated
for each sample position to create a 2D quality map that can guide the user in assess-
ing the quality/reliability of the elastogram shown (Figure 14.28).
SWE Artefacts 325
Mach 3 Mach 4
cone cone
2 ms–1
2 ms–1
6 ms–1 8 ms–1
FIGURE 14.27 Diagram showing the change in Mach cone angle as the Mach number increases
from 3 to 4.
FIGURE 14.28 SWE Elastogram of the breast on right with elastogram quality indicated on the
left. Stiffness, measured as Youngs Modulus, at the two measurement sites shown has been derived
from a set of repeated measurements. Source: Illustration by courtesy of Koninklijke Philips N.V.
SWE ARTEFACTS
Bang Artefact
The appearance of a stiff area in the near field caused by pushing down on the skin
with the transducer creating pre-compression of the tissue (Figure 14.29). This may
be avoided by use of coupling gel and less pressure on the transducer probe.
FIGURE 14.29 Example of the bang artefact in a SWE image of a breast where too much
pressure is applied to the skin giving a falsely high value at the top of the elastogram.
Source: Barr [12]/Korean Society of Ultrasound in Medicine.
REFERENCES
1. Bamber, J., Cosgrove, D., Dietrich, C.F. et al. (2013). EFSUMB Guidelines and recommendations
on the clinical use of ultrasound elastography. Part 1: basic principles and technology. Ultraschall
in der Medizin 34: 169–184.
2. Shiina, T., Nightingale, K.R., Palmeri, M.L. et al. (2015). WFUMB guidelines and recommendations
on the clinical use of ultrasound elastography: Part 1: basic principles and terminology. Ultrasound
in Medicine and Biology 14: 1126–1147. https://doi.org/10.1016/j.ultrasmedbio.2015.03.009.
3. Dietrich, C.F., Barr, R.G., Farrokh, A. et al. (2017). Strain elastography -how to do it? Ultrasound
International Open 3: E137–E149.
4. Barr, R.G. and Zhang, Z. (2012). Effects of precompression on elasticity imaging of the breast.
Journal of Ultrasound in Medicine 31: 895–902.
5. Barr, R.G. (2019). Future of breast elastography. Ultrasonography 38: 93–105. https://doi.
org/10.14366/usg.18053.
6. Dietrich, C.F., Barr, R.G., Farrokh, A. et al. (2017). Strain elastography -how to do it? Ultrasound
International Open 3: E137–E149. https://doi.org/10.1055/s-0043-119412.
7. Nightingale, K. (2011). Acoustic radiation force impulse (ARFI) imaging: a review. Current
Medical Imaging Reviews 7 (4): 328–339. https://doi.org/10.2174/157340511798038657.
8. Pereira, C.L.D., Santos, J.C., Arruda, R.M. et al. (2021). Evaluation of schistosomiasis mansoni
morbidity by hepatic and splenic elastography. Ultrasound in Medicine and Biology 47: 1235–
1243. https://doi.org/10.1016/j.ultrasmedbio.2021.01.022.
9. Fink, M. and Tanter, M. (2010). Multiwave imaging and super resolution. Physics Today 63: 28–
33. https://doi.org/10.1063/1.3326986.
References 327
10. Suh, C.H., Yoon, H.M., Jung, S.C. et al. (2019). Accuracy and precision of ultrasound shear wave
elasticity measurements according to target elasticity and acquisition depth: a phantom study.
PLoS One 14: e0219621. https://doi.org/10.1371/journal.pone.0219621.
11. Yoo, J., Lee, J.M., Joo, I. et al. (2020). Assessment of liver fibrosis using 2-dimensional shear
wave elastography: a prospective study of intra-and interobserver repeatability and compari-
son with point shear wave elastography. Ultrasonography 39: 52–59. https://doi.org/10.14366/
usg.19013.
12. Barr, R.G. (2019). Future of breast elastography. Ultrasonography 38: 93–105.
Appendix 1: Knobology
‘Knobology’ is the term that is frequently used to describe the operation of a scanners
controls in order to obtain the optimal image in an ultrasound examination. There are
standard settings that would generally be applied when imaging a particular area, for
example, abdomen, neck, muscular-skeletal, and when using particular modalities,
for example, B-mode, Doppler, elastography. In addition to the standard settings, con-
trols need to be adjusted throughout an examination to optimise the image and make
measurements.
Modern scanners have presets which set the scanner up for a particular examina-
tion using standard settings. These alter a large number of the controls. In addition,
techniques to automatically assess the image and then optimise it during the course of
an exam are being introduced. Some of the more portable scanners designed for point-
of-care use have very limited user access to controls. Presets and auto-adjustments
often do a very good job of managing image quality. However, it is important that
you as the sonographer are aware of what is being done and that you know how to
manually adjust the scanner to give the best results you can, for the patient in front
of you. There will always be situations where a good image is hard to obtain. Careful
adjustment of the controls is then needed to optimise the image to give the diagnostic
information required from the scan.
All of the controls and what they do have been described in this book. In this
appendix, they are presented in the form of a checklist, much in the way an aircraft
pilot will have a checklist to ensure their controls are working and in the correct posi-
tion for safe flight.
PRE-SCAN
• Is the room lighting suitable for scanning and viewing images?

• Is the monitor correctly set up for scanning so that it shows the full range of
greyscale?
• Is the patient couch and sonographer seat adjustable and set up for safe
scanning to avoid repetitive strain injury (RSI) problems for the sonographer?

329
330 Appendix 1: Knobology
B-MODE SCANNING
• Choice of probe – can you justify the choice of ultrasound probe for the
current exam?
• Probe type
• Choice of frequency for resolution vs penetration
• Is the probe clean?
Pre-Scan
• Are the presets set for the current exam?
• Is the correct Thermal Index (TI) selected for the current exam, e.g.
TIS, TIB, TIC?
• Is transmit power adjusted to a safe level whilst able to achieve adequate
penetration?
• Are there special factors such as scanning the eye, intraoperative or endo-
scopic exam of early pregnancy to be taken into account?
• Are the TI and MI values displayed on the screen showing safe operating
levels?
• Are the greyscale mapping (transfer function) and dynamic range set
appropriately?
• Are you using
• Compound scanning
• Adaptive processing
• Harmonic imaging
• Can you justify and explain their use?
During the Scan

• Adjust the gain and time gain compensation(TGC) to show a uniform image
at all depths
• Focus
• Set the focus to the depth of the target of interest
• Are you using multiple transmit focuses?
• Frame rate
• Is the frame rate adequate?
(affected by multiple transmit focuses, field of view size)
• Frame averaging – adjust to reduce blurring or to improve image quality
Making Measurements
• Correct choice of measurement and correct placement of calipers.
• Does the measurement need to be repeated?
CONTRAST-ENHANCED ULTRASOUND (CEUS)
• Awareness of sensitive tissues and increased risk of cavitation (cf MI)

• Use of harmonic imaging
DOPPLER WAVEFORMS
Pre-Scan
• Is the correct Thermal Index (TI) selected for the current exam, e.g. TIS, TIB, TIC?
• Is Doppler transmit power adjusted to a safe level whilst able to achieve ade-
quate penetration?
• Are the TI and MI values displayed on the screen showing safe operating levels?
During the Exam

• Is the sample volume length and position set correctly on the image?
• Can you justify the choice?
• Doppler angle
• Is the vessel correctly aligned?
• Is the angle cursor correctly aligned to the vessel?
• Is ‘heel and toeing’ required to achieve an adequate angle
• Can you justify the angle used?
• Is the wall filter setting adequate?
• Is the image frozen whilst the Doppler waveform is acquired?
• Is the Doppler gain set to show full black to white on the sonogram with no
over or under gain?
• Does the waveform need inverting?
• Is the scale set correctly so that the waveform fills the displayed range
without aliasing?
• Is the sweep speed (time base) set correctly to show the relevant number of
waveforms and no more?
332 Appendix 1: Knobology
Measurements
• Is autotrace or automatic caliper placement being used?
• Are the measurements being presented consistent with the observed wave-
form? (e.g. autotrace is not picking up noise, venous signal)
• Should manual placement of the cursors be used?
• Is a repeated measurement needed?
COLOUR DOPPLER ULTRASOUND (CDU)
Pre-Scan
• Is the correct Thermal Index (TI) selected for the current exam, e.g.
TIS, TIB, TIC?
• Is CDU transmit power adjusted to a safe level whilst able to achieve ade-
quate penetration?
During the Scan

• Does the scale need to be inverted to show the target of interest correctly?
• Adjust the colour gain so that colour is shown without colour bleeding
or dropout
• Is the wall filter setting adequate?
• Is the B-mode priority adjusted correctly?
• Frame Rate
• Adjust the size of the colour box to give an adequate frame rate
• Adjust the colour frame averaging to avoid blurring whilst improving
image quality
• Adjust the scale to avoid aliasing (or to use it diagnostically)
• Is ‘heel and toeing required to achieve an adequate angle or avoided to increase
sensitivity?
ELASTOGRAPHY
Pre-Scan
• Is the correct Thermal Index (TI) selected for the current exam, e.g. TIS, TIB, TIC?
During the Scan

Strain Elastography
• Is the elastogram ultrasound frequency set for adequate penetration vs
resolution?
• Set colour map for red/blue vs hard/soft
• Frame Rate
• Adjust frame rate control
• Lower elastogram ultrasound frequency lowers frame rate
• Reduce line density increases frame rate
• Adjust tracking window for sensitivity (improved SNR) vs resolution
• Adjust frame reject to balance noisy frames vs elastogram quality (affects
flashing of elastogram)
• Adjust noise reject to balance noise vs pixel dropout
• Adjust frame averaging to improve image quality and reduce flashing
• Adjust dynamic range to determine level of significant colour change shown
• Adjust colour blend for transparency of elastogram over B-mode display
Shear Wave Elastography
• Be aware of TI and MI values exceeding safe limits.

Appendix 2: Handling
Equations and Decibels
EQUATIONS
Equations may seem daunting and using them to calculate values not very familiar.
Their use within this book has been kept to a minimum, but it is not really possible to
grasp the essentials of ultrasound technology without including some equations and
showing some numerical results. This appendix has been included as a reminder of
how to manipulate equations to get the results we want to know.
It may be helpful to think of an equation as a special kind of diagram, composed
of numbers, or symbols representing numbers, that shows the relationship of one
item to another. The symbols often represent both a number and a property in the
real world.
For example velocity v0 = 5v1
The property is the velocity of something and this equation tells us that whatever
value v0 has, v1 will have 5 times that value.
Each side of an equation has a top and a bottom, but where the bottom equals 1,
we do not usually write it.
v AB D F D
Examples: v or normally written AB F
1 1 E 1 E
An equation is like a ‘seesaw’ type of weighing scales where one side must always
balance or equal the other. In order to make sure you have the right result you can
write out every stage in manipulating the equation, but there are a number of short
cuts (rules) you can use as shown in the examples below.
Rule 0 You can change things around on an equation but whatever you do to one
side you must also do to the other, so as to keep it in balance.

335
336 Appendix 2: Handling Equations and Decibels
Adding or Subtracting A = B subtract B from A ⟹ A − B = B − B = 0
B on both sides
i.e. A − B = 0 and note how B has changed sides. This gives:
Rule 1 If an item moves from one side to the other it changes sign.
Dividing A B
A B 1 (divide both sides by B) This gives:
B B
Rule 2 If the same item appears on the top and the bottom of one side, we can
cancel them out and replace them with 1.
and
Rule 3 We can move an item from the top of one side to the bottom of the other side
(and vice versa) and the equation stays in balance.
Example A C AD CD
C (D moves from the bottom to the top)
B D B D
Multiplying Top and bottom of one side by the same amount.
BB B2
Example A B A
B B
ere we could cancel the B’s on the right to get back to where we started, but
H
it is sometimes useful when rearranging equations to do the same thing to
the top and bottom of one side and leave it there before doing something else.
This gives:
Rule 4 You can do the same thing to the top and the bottom of one side and the
equation will stay in balance.
Multiplying and dividing when there are several terms
Example A + B = C ⟹ A. D + B. D = C. D = D(A + B)
D was a new item but we multiplied it to all terms on both sides so the equation
stays in balance. This gives:
Rule 5 Where there are several terms (separated by + or −) on one side, then you
must do the same to each of the terms.
and
Rule 6 Where the same item is in each term then the terms can be collected together
in parentheses with the common item outside. This also means that to
expand such parentheses, each term inside the parentheses is multiplied by
each item outside the parentheses.
A B C A B
Example A B C 1
C C C C
A and B terms are each divided by the same amount C. This gives:
Rule 7 If all terms on one side are divided by the same item, we can collect them
together over that item (but only if they have the same item on the bottom).
A B A B B
Example
C CD C D
i.e. WRONG because on the bottom of the left side not all terms are the same.
A B AD B AD B
However
C CD CD CD CD
by multiplying top and bottom of the first term by D all terms then have CD
on the bottom.
Rule 8 Anything in parentheses can be treated as a single item.
E
Examples E D A C D
A C
parentheses moved as per Rule 3
AB E
E D D
C AB
C
AB
Here has been treated as a single item and moved to the bottom of the
C
other side as per Rule 3. Then:
1
Rule 9 is the same as inverting the fraction on the bottom and putting it in
fraction
the normal position
1 B B
Example
A AB A
B B
here we multiply top and bottom by B and the B’s on the bottom then cancel.
Tip If you want to check that you have done the right thing in manipulating an
equation, then put in some numbers and see if it gives the right answer.
A B 2 3
Example A 2, B 3, C 4, D 5 0.65
C CD 4 20
AD B 10 3
rearrange the equation 0.65
CD 20
Doppler Equation as an Example

The equation may be remembered as shown on the left. It gives the Doppler frequency
from a target velocity v, speed of sound c, the transmitted frequency fT tissue and the
Doppler angle θD.
v fD c
fD 2 f cos V
c T D
fT 2 cos D
The equation shows the relationship between these quantities and we see that for
a given transmit frequency the Doppler frequency depends on the ratio of the target
velocity to the speed of sound in tissue. It also changes as the cosine of the Doppler
angle. We know that cos 90° = 0, so a Doppler angle of 90° gives a Doppler frequency
of 0 regardless of the target velocity.
The thing we want to know is the target velocity, say, blood velocity, in the body.
On the right, the Doppler equation has been rearranged using Rule 3 to show the
relationships in terms of the target velocity. Once we know the Doppler frequency for
a given angle, the blood velocity producing that frequency can be calculated and we
see that the value depends on the ratio of the Doppler frequency to the transmitted
frequency.
Tips for Calculating Answers to Problems

1. Make sure all of the units are consistent
e.g. kilograms (kg), metres (m), seconds (s)
Do not mix, for example, millimetres with metres in the calculation or you
will end up with errors in factors of 10.
2. Make sure you have calculated all the factors of 10 correctly

Factors of 10 can often be the source of errors in the answer. Multiplying a
number by a power of 10 is the same as moving the decimal point that number
of places to the right, and dividing by a power of 10 is the same as moving the
decimal place to the left.
Examples 15.72 100 1572.0
this can also be written as 1.572 × 103 that is, as a unitary number multiplied
by a power of 10.
5.78/1000 = 0.00587 = 5.78 × 103
0.0733 = 73.3 × 103
Note:
• Anything to the power of 1 is itself, e.g. 41 = 4
• Anything to the power of 0 is 1, e.g. 100 = 1, 60 = 1
To multiply powers of 10 add the indices
e.g. 106 10 4 1010
To divide powers of 10 subtract the indices
102 3 1
e.g. 10
105 103
Note:
• it is always ‘top minus bottom’
• a power can move from top to bottom with a change in sign
1
e.g. 10 2
0.01 1 10 2
but we usually omit the ‘1’
100
TABLE A2.1 Units used in expressing the values of variables.

Power Name Symbol
106 mega M
103
kilo k
101
10
100
1
10−2
centi c
10−3
milli m
10−6
micro μ
10−9
nano n
To make sure you have calculated the powers of 10 correctly, it is a good idea to
write all the numbers as unitary numbers with powers of 10 and then calculate the
result from the numbers and then add up the powers of 10.
500 0.0072 5 102 7.2 10 3

4
e.g. 6 10
60 10 4 0.01 6 105 10 2
if this was distance in metres then the answer in millimetres = 0.6 mm, i.e.
divided by 1000
3. Ask yourself the question ‘is this a sensible answer?’
e.g. blood velocity = 500 ms−1 probably wrong by a factor or 100 or 1000!
4. To get the units (e.g. kg, m, s) right, do the calculation of the numbers. Then
write the equation with just the units and rearrange the equation to see in what
units the numerical answer should be given.
Example: In Chapter 1 we saw that the speed of sound in a medium was
given by the square root of the bulk modulus κ divided by the density ρ
P
Where K V has units kgm−1s−2 and density ρ has units kgm−3
V
K kg m 1s 2
Then, looking at the units c 3

m2s 2
ms 1
kgm
Which is the correct units for a speed of sound. Note how the units in an
equation behave in the same way as any symbols or numbers.
Final Example Combining All These Rules and Tips

Question
If the speed of sound in a plastic is 4.7 kms−1 and a reflector is 3 cm away from the
transducer, what is the pulse-echo time from the reflector?
Answer
1. Find the right equation (draw a diagram if it helps, Figure A2.1 shows the
pulse-echo principle with the variables to be determined indicated)
d
c
Echo
FIGURE A2.1 The pulse-echo principle.

distance
velocity (think of speed = miles per hour) so the equation we need is
time
2d
c where the 2 is for go and return distance 2d.
t
2. Rearrange the equation to get it in the right form to give the answer
2d
t
c
3. Check all the units are consistent

c = 4.7 kms−1 = 4.7 × 103 ms−1
d = 3 cm = 3 × 10−2m
everything is now in metres and seconds
4. Do the calculation
2 3 10 2
t 1.28 10 5 s 12.8 s
4.7 103
5. Is this a sensible answer?

Recall that for soft tissue the average speed of sound is 1540 ms−1 and
go and return time is 13 μs per centimetre. Here the speed of sound is about
3 × 1540 and the distance is 3 cm so 12.8 μs is the right answer.
DECIBEL NOTATION
We often need to measure the ratio of change in energy, power or intensity
e.g. -proportion of energy reflected at an interface

- amount absorbed
- gain of an amplifier e.g. 1 : 1000
There is a compact way of writing and calculating such ratios that avoids handling
large numbers. To do this we take the logarithm of the ratio. This then gives the
change as Decibels (dB)
defined as: number of dB = 10. log10 (ratio)

e.g.: 10. log10 (2/1) = +3dB
a ratio of a doubling, i.e. 2 : 1 has a dB value very close to 3dB
conversely: ratio = antilog10 (no. of dB/10) which for +3dB gives a ratio of 2
Advantages
• whereas ratios multiply or divide, decibels add or subtract
1 1 1
e.g. multiply but add 3dB 3dB 6dB
2 2 4
• it is easy to express very large ratios as small numbers.
e.g. 10 000 : 1 40dB
So: ratio 2 corresponds to +3dB

ratio10 corresponds to +10dB
ratio 0.5 corresponds to −3dB
ratio 0.1 corresponds to −10dB
NOTE
Where the ratio is fractional, e.g. 0.5, the decibel value is negative.
Now we can easily convert a ratio using multiples of 2 and 10
e.g. 1 : 4000 2 2 10 10 10
3 3 10 10 10 36dB
Conversely −60dB = −30−30 = 10−3× 10−3 = 10−6
Note the similarity to adding indices for multiples of 10 or using a logarithmic

scale on a graph to show a very large range of values.
or by direct calculation
e.g. 15% 0.15 10. log10 0.15 8.24dB

IMPORTANT NOTE
Sometimes dB are used to express the ratio of two pressures or voltages. These are
values of the amplitude of the signal. Since power or intensity is ∝ (amplitude)2
i.e. p2 or V 2, then for expressing amplitudes as dB we must use:
dB 20 log10 amplitude so a doubling of amplitude 6dB

a ratio of 10 in amplitude 20dB, etc.
Appendix 3: The Unfocused
Transducer Beam Shape
If we look at a field point in the ultrasound beam, we can measure the geometrical
distance to various points across the transducer aperture to determine whether there
will be constructive or destructive interference at that field point. The easiest way to
think about it is to do the reverse, to consider an echo arising from the field point and
to look at what the transducer ‘sees’ of the phase from that echo. In this Appendix,
we consider a plane unfocused a transducer with continuous wave transmission at a
single frequency and consider the receive ultrasound beam, i.e. the one-way beam.
The transmit beam will have the same shape and the go-and-return beam shape will
be the combination of the two beams. Several field points that we will consider are
numbered on the Figure A3.1.
4
3 2 1
5
6
FIGURE A3.1 Intensity contour map of a continuous wave ultrasound beam from a plane
aperture. Index points indicated.
A POINT IN THE FAR FIELD (FIGURE A3.2)
In the far field, field point (1) – beyond the last axial maximum – the difference in
path length between the centre of the aperture and the edge of the aperture is less
than a wavelength λ and the wavefronts approximate plane waves. Another way of

345
346 Appendix 3: The Unfocused Transducer Beam Shape
FIGURE A3.2 Wavefronts arriving at the transducer from a point in the far field.
putting it is to say that the whole of the aperture ‘sees’ the same phase of the wave-
front at the same time from that field point.
LAST AXIAL MAXIMUM (FIGURE A3.3)
At field point (2), the edge of the aperture has a path length of just λ/2 different
to the centre of the aperture. This means each half of the aperture ‘sees’ a positive
phase at time t0, so every point across the aperture contributes positively to the net
amplitude seen at (2). Half a cycle later, each half of the aperture will see a
negative going phase. As the phase of the echo across the whole aperture adds up
(constructive interference), this gives a strong intensity at field point (2), i.e. the last
axial maximum.
λ
2
t0
FIGURE A3.3 Wavefronts arriving at the transducer from the last axial maximum.
Appendix 3: The Unfocused Transducer Beam Shape 347
POINTS IN THE NEAR FIELD (FIGURE A3.4)
At field point (3), on the central axis of the beam, the path length between the centre
and edge of the aperture has a phase difference of one whole wavelength. So each half
of the aperture ‘sees’ equal amounts of positive and negative phase. These cancel out
(destructive interference) and the intensity at (3) is zero.
At field point (4) each half of the half-aperture ‘sees’ a positive phase so the inten-
sity at (4) is relatively high. This gives one of the peaks seen in the near field.
Going through this process for all points in the beam shows how the complex
pattern of intensity in the near field arises.
λ
2
FIGURE A3.4 Wavefronts arriving at the transducer from a point in the near field (3) and field
point (4).
EDGE OF THE BEAM (FIGURE A3.5)
At field point (5) each half of the aperture ‘sees’ the opposite sign of phase, so across
the full width there is destructive interference giving a result of zero intensity, i.e. the
edge of beam.
NOTE
The edge of beam (5) is not on the line of the geometrical parallel beam seen in the
first simplified beam shape diagram (Figure 3.2), but is towards the centre-line of
the beam. In other words, diffraction gives a natural focusing to the beam at the
last axial maximum and the beam there is ~25% narrower than the parallel beam
approximation.
348 Appendix 3: The Unfocused Transducer Beam Shape
λ
2
λ
2
FIGURE A3.5 Wavefronts arriving at the transducer from the edge of the beam.
SIDE LOBE (FIGURE A3.6)
At field point (6), each half of the aperture again ‘sees’ a net positive phase so there
is an increase in intensity, giving the first side lobe. However, because only 1/3 of
the aperture is giving a positive contribution (2/3 cancels out), the intensity is much
less than that of the main lobe and the intensity of the first side lobe is -18dB below
main lobe.
2λ
3
2λ
3
FIGURE A3.6 Wavefronts arriving at the transducer from the first sidelobe.
SUMMARY
What we see described in the intensity contour map is the diffraction pattern of a
plane transducer with plane waves of a single frequency. A real beam plot of a such a
transducer is seen in Figure 3.4.
Index
A Attenuation, 34–37
A/B ratio, 221 Attenuation coefficient, 35, 231
Absolute risk, 233 Autocorrelation, 176–177
Absorption, 34, 231 Autotrace, 166–170
Absorption coefficient, 231 Average velocity, 167–170, 222
Accuracy, 204 Axial misplacement, 123
Acoustic impedance, 20, 25 Axial resolution, 55, 208
Acoustic lens, 44, 47
Acoustic pressure, 9 B
Acoustic radiation force imaging, 314–315 Backing layer, 62
Acoustic window, 111 Backscatter, 251
Adaptive beamforming, 285 Ball and spring model, 6, 16
Adaptive filtering, 118 Bandwidth, 14–16, 57, 67, 90, 159, 169, 179
Air reverberation, 241 Bang artefact, 325
ALARA, 239 Bayesian statistics, 266
Aliasing, 154–158, 183–184 Beam divergence, 35
Ambient lighting, 121 Beamforming, 47, 71, 96, 97, 247, 277–283,
A‐mode, 74 285, 288, 292–294, 296, 322
Amplitude, 9, 101 Beam steering, 50
Amplitude modulation, 253 Bernoulli equation, 260
Analogue to digital conversion, 95–101 B‐flow, 256–260
Angle cursor, 145 Blue/green/red artefact, 314
Anisotropy, 36, 134, 231, 310 B‐mode, 76
Aperture, 39, 53, 278–279 B‐mode priority, 188
Apodization, 50 Bulk modulus, 18, 317
Area measurement, 213–215 Bulls eye plot, 201
Arrays, 68–71
Artefacts, 122–140, 170–173, 190–194, 283, C
310–314, 325 Caliper accuracy/precision, 207–208
Artificial intelligence, 261 Capacitative transducer, 66–68
Artificial neural networks, 266–268 Cavitation, 236–238

349
350 Index
Cavitation nuclei, 238 D

CEUS, 240, 249–256 Damping, 62
Chirp pulse, 94 Dark adaption, 121
Cine loop, 82 Decibels, 23, 341–343
Circumference measurement, 213–215 Deep learning, 261, 266
Classification, 265, 266 Deformation, 195
Clutter, 42, 58, 90, 136 Delay and sum beam forming, 51, 278
CMUT, 66–68 Demodulated echo, 101
Coded excitation, 92–94, 177, 258 Depth gain control, 98–99, 127, 128
Coherent imaging, 31, 115 Destructive interference, 44
Coherent spatial compounding, 282 Diffraction, 43
Collapse cavitation, 237 Diffraction limited resolution, 283
Colour aliasing, 183–184 Displacement, 195
Colour blend, 308 Distortion, 195
Colour box, 181 Doppler angle, 143, 162–164, 183, 217
Colour Doppler, 175, 221, 286 Doppler angle measurement, 216–223
Colour dropout, 187, 188, 190–192, 332 Doppler artefacts, 170
Colour enhancement, 120 Doppler effect, 141
Colour flow mapping, 175 Doppler equation, 143, 157
Colour map, 175, 308 Doppler frequency, 142, 180
Colour M‐mode, 194 Doppler frequency offset, 147,
Colour scale, 180, 185 151, 165, 180
Comet tail, 132 Doppler frequency resolution, 151
Compounding, 124, 282, 323 Doppler gain, 165
Compression, 102, 237 Doppler Pressures, 260, 291, 298
Compression speed, 308 Doppler pulses, 159
Computed ultrasound Doppler sample volume, 145, 146, 152,
tomography, 285 154, 159, 163, 219, 222
Computer aided diagnosis, 263 Doppler scale, 156, 158, 164
Constructive interference, 44 Doppler shift, 142
Continuous wave Doppler, 145–152 Doppler spectrum, 149–150, 160, 166
Contrast, 102 Duplex, 144, 154
Contrast agents, 240 Dwell time, 230, 233
Contrast enhanced ultrasound, 249–256 Dynamic focusing, 85
Contrast pulse sequencing, 255 Dynamic range, 102–106, 120, 308
Contrast ratio, 252
Contrast resolution, 106, 118–121 E
Contrast ultrasound, 240 Echo amplitude, 101
Correlation, 93, 176 Echocardiography, 186, 218, 260, 262, 266
Cross correlation, 93, 176 Echo detection, 101
C‐scan, 81 Edge enhancement, 118
Curvilinear array, 68, 77 Edge enhancement artefact, 312
CUTE, 285 Elasticity, 301
CW Doppler, 145–152 Elastogram, 306
Index 351
Elastogram box, 306, 309 Frequency bins, 149

Elastography, 303 Frequency domain, 14
Electronic focusing, 52 Frequency offset, 147, 151, 165, 180
Element dropout, 242 Frequency resolution, 151
Elevation plane, 47, 135 Fresnel zone, 40
Endoprobes, 82–83, 234, 241 Fundamental, 87
Energy, 11, 22, 225–226, 230, 232, 234, Fusion imaging, 269–271
235, 314, 321
Energy spectrum, 13 G
Ensemble length, 179, 181, 288 Gain, 97–99, 102, 120, 165
Envelope detection, 101 Gaussian beam, 50, 295
Errors, 206 Grating lobes, 48, 78, 136
Extended field of view, 78 Grey scale imaging, 76
Grey scale mapping, 102–106, 120
F Ground truth data, 268
Far field, 40, 345
Fast Fourier transform H
(FFT), 148 Hanafy lens, 68
Feature extraction, 265, 266 Hand held Doppler, 145
Filters, 90 Harmonic imaging, 87–92
First interface, 209 Harmonics, 87, 252
First moment, 167 Hazzard, 225
Flash artefact, 192 Heating, 231–234
Flashing, 240, 255, 307, 308 Heel and toeing, 183
Flow vector, 293 High pass filter, 90
Focal depth, 44, 53, 96 Huygen’s source, 43
Focal length, 44, 45 Huygen’s wavelets, 43
Focal strength, 46 Hydrophone, 227
Focal width, 45, 53
Focal zone, 45 I
Focus, 111, 113, 154, 161, 169, 222, Ideal source, 278
280–283, 292 Image contrast, 102
Focusing, 44, 52, 84–85 Image plane, 47, 76
Forward flow, 146 Image processing, 108, 261–269
4D scanning, 70, 82, 262 Image registration, 269–270
Fourier transform, 148 Inertial cavitation, 237
Frame averaging, 116, 182, 308 Instantaneous Doppler spectrum, 149, 159
Frame freeze, 106, 209 Intensity, 22, 26, 226–230, 240,
Frame rate, 77, 86, 112–115, 181, 281–282, 252, 283, 314
289, 307 Intensity contour map, 42, 227
Frame reject, 307 Interference, 43–44, 48, 294, 304
Fraunhofer zone, 40 Interlacing, 113
Free radicals, 238 Interobserver variability, 205
Frequency, 5 Interpolation, 114, 181, 321
352 Index
Intraobserver variability, 205 Noise rejection, 102

Intrinsic spectral broadening, 160– Non‐linear propagation, 89, 252
162, 219, 290 Non‐thermal effects, 235–239
Invert, 165 Normalised first moment, 167–168
ISATA, 230 Nyquist limit, 155, 183–185
ISPPA, 230
ISPTA, 228 O
ISTPT, 230 Offset frequency, 147, 151, 165, 180
1.5D array, 70
L OTTO, 244
Lagrangian strain, 197
Lateral resolution, 55, 208, 281 P
Lighting, 121 Panoramic view, 78
Linear array, 47, 76 Parabolic profile, 148
Linear propagation, 89 Peak systolic velocity, 162, 165, 167
Linear scan, 76 Peak velocity envelope, 165–167
Low pass filter, 90 Penetration, 31, 37, 57, 84, 309,
315, 320, 324
M Period, 7
Mach cone, 322 Personal systematic error, 211
Machine learning, 261 Phantoms, 120
Mach number, 323 Phase, 10, 31, 48, 50, 53, 96, 101, 115,
Mapping, 102–106, 175, 201, 285, 296, 176–180, 253–255, 277, 279, 282,
306, 308, 324 285, 294
Matching, 63–65 Phased array, 50, 52, 68, 78
Matching layer, 62 Piezoelectric effect, 59
Matrix array, 70, 80 Plane wave, 10, 39
Maximum velocity, 156 Plane wave beam forming, 278–283
Measurement protocols, 209, 212 Plug flow, 149
Measurements, 109 Point shear wave
Mechanical index, 239, 240 elastography, 320
Microvascular imaging, 289 Point source, 39, 278
Mirror image artefact, 130, 192 Poisson effect, 302
M‐mode, 74, 194 Poor visualisation, 29, 125
Monitor, 121 Post cystic enhancement, 128
Multizone focusing, 84 Pourcelot’s resistance index, 220
Power, 13, 22, 31, 37–38, 149, 165, 226–227,
N 232, 235, 239
Navier‐Stokes equation, 260, 298 Power Doppler, 188–190
Near field, 40, 347 Precision, 204
Needle guidance, 271–274 Pre‐compression, 310
Needle guide, 272 Pre‐loading, 310
Needle visualisation, 271–274 Pre‐processing, 265
Neural networks, 266–268 Presets, 106, 108–109, 194
Noise, 37, 68, 92, 98, 101, 102, 119 Pressure gradient, 260, 298
Index 353
Pressure measurement, 260, 298 Rectified diffusion, 236–237

Pressure wave, 9 Red/green/blue artefact, 314
PRF, 96, 153, 155–158, 185, 290 Reflection, 25, 130
Probe foot plate, 306 Refraction, 27, 122
Probe self‐heating, 62, 234 Registration, 269–270
Protocols, 209, 212 Relative risk, 233
PSV, 162, 165, 167 Repetitive stress injury, 245
Pulsatility index, 219 Reproducibility, 205
Pulse average intensity, 228 Resistance index, 220
Pulse compression, 92 Resolution, 54–57, 84, 86, 89, 111, 112, 124,
Pulse‐echo principle, 73 208, 212, 244, 283, 296
Pulse envelope, 13 Resonance, 252
Pulse inversion, 91, 254 Retrograde flow, 186
Pulse repetition frequency, 96, 153, Reverberation, 131–134
155–158, 185, 290 Reverse flow, 146, 186
Pulses, 11, 31, 33, 37, 54–57, 62, 73, RF signal, 101, 277
92, 95–96, 227–230, 238, 240, 255, Rigidity, 301
278, 314–315 Ringdown, 131
Pulse wave Doppler, 152–159, 290 Risk, 225
Push pulse, 314 Rouleaux, 256
PVDF, 60 Routine checks, 241–244
PW Doppler, 152–159, 290
PZT, 50 S
Sample and hold, 154
Q Sample volume, 144, 146, 152, 219
Quality assurance, 241–245 Sampling, 153, 182
Quality factor, 16 Scale, Doppler, 156, 158, 164, 180, 185, 216
Quality indicator, 309 Scaling, 207, 269
Quantisation error, 208, 217 Scan converter, 106
Quarter wave matching layer, 62 Scan plane, 47, 76
Scattering, 30, 35, 56, 251
R Sector scan, 76
Radial image, 82 Segmentation, 262, 265
Radiant flow, 190–191 Segmentation seed, 262
Radiation force, 235, 314 Self heating, 62, 234
Radiation force balance, 235 Shadowing, 127, 173, 190, 266, 314
Random error, 206 Shear modulus, 318
Range gate, 152 Shear wave, 316
Rarefaction, 237–238, 240 Shear wave artefacts, 325
Rayleigh scattering, 30, 56 Shear wave compounding, 323
Read zoom, 107 Shear wave elastography, 316–321
Real time, 47, 77, 148, 269, 273, Side lobes, 42, 136, 348
277, 296, 322 Signal compression, 102
Receive focus, 84, 281 Signal to noise ratio, 92, 199, 283, 288, 305
Rectification, 101 Single crystal, 61
354 Index
Slice thickness, 47, 135 Temporal peak intensity, 228

Slip plane, 311–312 Temporal resolution, 112, 117, 288
Snell’s law, 28 Test objects, 244
Sonogram, 148–152 TGC, 98
Sound wave, 5 Thermal effects, 231–234
Sparse array, 71, 278 Thermal index (TI), 232
Spatial average intensity, 228 3D imaging, 80–82, 190, 262–264, 266–267,
Spatial compounding, 117, 122, 282 269–270, 274, 297
Spatial peak intensity, 228 Thresholds, 102
Speckle, 31, 56, 115–117 Time domain, 14
Speckle tracking, 176, 304 Time gain compensation, 98–99, 127, 128
Speckle tracking echocardiogra- Tissue Doppler imaging, 194–199
phy, 199–202 Tissue heating, 231–234
Spectrum, 13, 149 Tissue mimicking phantom, 120
Specular reflection, 25 Tissue sensitivity, 230
Speed of sound, 6–8, 16–20, 122, 143, 207, Tracking probe position, 269
244, 283, 317–320, 323 Tracking window, 305, 307
Speed of sound artefacts, 122 Transducer, 9, 40, 59–62
Speed of sound correction, 122, 283–286 Transducer self‐heating, 62, 234
Stable cavitation, 236 Transmit, 95
Stiffness, 301 Transmit focus, 84, 281
Strain, 302 Transmit power, 37, 96, 165, 230, 241, 293
Strain elastography, 303–314 Transverse oscillation, 293–296
Strain elastography artefacts, 310–314 Transverse wave, 316
Strain estimating window, 305 Twinkle artefact, 192
Strain imaging, 197–202 2D array, 70
Strain rate, 197
Strain ratio, 316 U
Streaming, 235 UFCD, 286
Strength of focus, 46 Ultrafast ultrasound, 277
Stress, 502 Ultrafine measurements, 212
Stress concentration artefact, 312 Ultrasound, 5
Superposition, 32–33, 43, 87–88, 115 Ultrasound tomography, 285
Sweep speed, 164, 216
Swept gain, 98–99, 127, 128 V
Synthetic aperture, 113, 278 Vector flow imaging, 291–298
Synthetic aperture imaging, 86, 278 Vector flow map, 293
Systematic error, 206 Velocity profile, 148–150
System velocity, 73, 124, 207, 283 Video pulse, 101
Viewing monitor, 121
T Visualisation, 29, 125, 209
Target visualisation, 29, 125, 209 Visual perception, 120
Temporal artefacts, 137–139, 192 Volume flow, 221–223, 298
Temporal average intensity, 228 Volume measurement, 213–215
Index 355
W Wideband transducer, 16, 65

Wall filter, 171–172, 187–188 Work related stress injury, 245
Wall thump, 171 Wrap around, 156, 184
Wave amplitude, 9 Write zoom, 108
Waveform ghosting, 172
Waveform indices, 219–221 Y
Wavefront, 10 Youngs modulus, 302, 319, 320
Wavelength, 8
Weighted average velocity, 167 Z
Whisp of smoke, 256 Zoom, 107–108, 121
WILEY END USER LICENSE AGREEMENT
Go to www.wiley.com/go/eula to access Wiley’s ebook EULA.

Ultrasound Technology For Clinical Practitioners

Uploaded by

Copyright:

Available Formats

Ultrasound Technology For Clinical Practitioners

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ultrasound Technology For Clinical Practitioners

Uploaded by

Copyright:

Available Formats

Ultrasound Technology

for Clinical Practitioners

Limit of Liability/Disclaimer of Warranty

Library of Congress Cataloging-­in-­Publication Data Applied for:

Cover Image: Courtesy of the Author

Set in 10.5/13pt STIXTwoText by Straive, Pondicherry, India

CHAPTER 1 The Basic Physics of Ultrasound 5

CHAPTER 2 The Interaction of Ultrasound with Tissue 25

CHAPTER 3 Beam Shapes 39

CHAPTER 4 The Ultrasound Probe 59

CHAPTER 5 Image Formation 73

CHAPTER 6 The B-­Mode Scanner 95

CHAPTER 7 Image Quality and Artefacts 111

CHAPTER 8 Principles of Doppler ­Ultrasound 141

CHAPTER 9 Principles of Colour Doppler Ultrasound 175

CDU and the Doppler Angle 183

CHAPTER 10 Making Measurements 203

Measurements Protocols 209

CHAPTER 11 Safety and Quality Assurance 225

Suggested Routine User Checks 241

CHAPTER 12 Advanced Topics 249

CHAPTER 13 Ultrafast Ultrasound 277

Transverse Oscillation (TO) 293

CHAPTER 14 Elastography 301

Appendix 1: Knobology 329

ADC analogue to digital converter

ISPTA spatial peak temporal average intensity

TIS thermal index for soft tissue

Ultrasound Technology for Clinical Practitioners, First Edition. Crispian Oates.

sonographer needs anatomical knowledge and an understanding of how the images

A sound wave is a fluctuating variation in pressure within a medium such as air,

Ultrasound Technology for Clinical Practitioners, First Edition. Crispian Oates.

THE SPEED OF SOUND EQUATION

Relationship Between Pressure, Particle Velocity and Particle

FIGURE 1.6 Definition of wave amplitude A.

180° 270° 360° = 0°

FIGURE 1.8 Definition of the phase angle of a sound wave.

90 180 270 360

FIGURE 1.9 Two sine waves with a phase difference of 45°.

ENERGY IN A SOUND WAVE

So, for the acoustic pressure p, energy E p2 .

• Peak amplitude – typically, 1 atmosphere = 105 Pa

• Pulse length in tissue – three—­four cycles

FIGURE 1.12 Typical pulse shape of a short pulse of ultrasound.

FIGURE 1.13 The pulse envelope of an ultrasound pulse in space.

ENERGY SPECTRUM OF A PULSE

• For example, in a light spectrum, as seen in a rainbow, red has a lower

The rate at which energy is produced or transmitted is called power, which is

It has all of its energy (E0) at a single frequency ( f0 ).

• What is the frequency of a pulse?

FIGURE 1.15 A 7-­cycle pulse.

Time domain Frequency domain

E(f) Short pulse

Infinitesimally short pulse

The importance of this relationship for medical ultrasound is that in order to

An approximation of the bandwidth of a given pulse length is given by

So, a wideband transducer has a low Q, and a transducer with a narrow

SPEED OF SOUND (C)

(Note: do not confuse the speed of sound with particle velocity!)

Library of Congress Cataloging-in-Publication Data Applied for:

CHAPTER 6 The B-Mode Scanner 95

CHAPTER 8 Principles of Doppler Ultrasound 141

• Pulse length in tissue – three—four cycles

FIGURE 1.15 A 7-cycle pulse.

different to their cross-sectional structure. Their scattering and reflectivity, therefore,

In order to have narrow beams, imaging ultrasound uses high-frequency ultra-